Mother's Day Gift Guide, Rorra in WSJ, Morgan Housel, Mehul Nariyawala, Sonya Huang, Stanford Review, Will Quist, Aidan Dewar, George Maloney

30:55At first, the railroad wasn't a priority. When David had played with toy trains growing up, he said he wouldn't describe himself as a train guy. but a friend offered to introduce David to the late Bill Tyson, a one-time imagineer and garden railroad enthusiast who could help restore it. Then they visited the South Coast Railroad Museum near Santa Barbara where they rode a garden railroad for$1. At the end of it, we said, okay, we have to do this. It's just too much fun. It was the most expensive dollar I ever spent. That's hilarious. It's great. That's great. We have another property. Another property.

31:27This This is another good option for Mother's Day if you're in the market to buy your mother a house. She's trying to move south, closer to the tropics. Yeah. Why not Fort Lauderdale? Better weather, Fort Lauderdale. There's a new massive waterfront home. 15 ,000 square feet. $39 million. And it includes eight separate bar areas. So for the entertainer. So a heavy... I mean, people are saying alcohol is coming back. There's so much alpha with alcohol being out of style. Maybe go long alcohol. and enjoy the eight bars. But you can also serve mocktails and beverages for kids. You can do anything.

32:03Energy drinks. Energy drinks. Yeah. For more than 20 years, Steve Savor has been hosting lavish parties at his waterfront mansion in Fort Lauderdale, Florida. Last year, he hosted a Barbie-themed gala, decking out the house in hot pink and dressing up as Ken. I always say, if the house burned to the ground and I didn't have any insurance, I got my money's worth out of it. The 64-year-old retired bachelor. That's hilarious. he's now listing the property known as villa de palma for 39 million dollars he says he travels frequently especially during the summer months and wants a home that doesn't require as much attention so the house is saver who was trained as a lawyer made his fortune in merchant banking and is a former ceo of the pittsburgh-based communications company comnet erickson retiring at age 39 he assembled villa de palma over two decades starting with the purchase of the main house for about 1.8 million in 2002.

32:53Property records show Florida is known for massive fluctuation in their housing market. In the years since he has spent millions adding more land to expand his footprint to roughly an acre. One of those properties included a five bedroom house, which he raised around 2009. This guy just turned making this party house into his full-time job. Raised means demolished, by the way. And so he says to celebrate the demolition, he hosted an animal house style toga party. He spent two weeks decorating the house to look like a fraternity house. Then the night before the demolition, he and his pals tore up the house with sledgehammers.

33:32Can you imagine that? At the end of the night, we were throwing kegs through windows, he said. What a bro. This is so insane. After tearing down the small house, Savore has expanded the main house, completing the project around 2010, The eight-bedroom Mediterranean-style estate has about 15 ,000 square feet of living space and two large outdoor pools. It's well-equipped for entertaining with a wine cellar and eight separate bar areas, including one inspired by Hotel du Cap Eden Roque in the south of France. Fantastic hotel. Have you been? No. And Savor calls another bar on the rooftop Tequila Tower.

34:10This guy is such a party animal. A giant table in the space can hold plenty of people to dance on after tequila has been depleted. Savage. There are two relaxation rooms, properties within walking distance of the beach in downtown Fort Lauderdale's main shopping strip. I would come down for long three-day weekends and ended up staying two weeks. When I retired, I made my playground my home. What a legend. Anyway, great option. I'm interested to see where this one lands. Yeah. Anyway, we have two minutes until our first guest is joining. Let's do some timeline. What else is in the news? Actually, let's do an ad.

34:45Eight Sleep. what's new pod 4 ultra pod 4 has all the signature features you love about the pod plus new groundbreaking upgrades 5 year warranty 30 night risk free trial free returns free shipping you gotta love it I went hard last night I did terribly I got 85 96 96 it's rough 93 minutes of deep sleep one last potential gift for mothers on mother's day a billboard go to adquip.com they make out of home advertising easy and measurable, say goodbye to the headaches of out-of-home advertising, and buy a massive out-of-home campaign. Huge opportunity to just buy the 10 billboards closest to your mother's home, wherever she lives, and make it so that she can't leave her house without seeing how much you love her.

35:34Exactly. So we got a post here from, we'll cover this, and then we'll go into the guest from Bucko Capital Bloke. He posted back on March 3rd. New Mag 7. Reinmittal, which is German arms manufacturer. Solana. VCS coin. Palantir. Salesforce for... Okay. He's not pulling punches. Hems, DTC. I can't say anything. Bloke. All new appliances and vehicles purchased before the tariffs. gold and so he then quotes it and by saying uh ryan mattal is up 54 solana's up 15 palantir up 42 percent hymns up 26 dave up 60 used vehicles up 10 gold 14 and the max seven is down three percent i wonder if this is actually just over the last two months maybe yeah i guess this is wait do the tariffs hit mark's third something like that um anyway congrats to him for calling it clean up the language if you want to be on the show more uh anyway uh they said to cheat on everything so i decided to cheat their company i generated 84 000 believable resumes to spam their job applicants and sent ai agents to waste their time i'm shorting the vcs defending them you see this fight that's going on can we michael are we able to pull this up on the screen so rise is trolling roy former guest on the show so yeah roy has a company called clueley he is helping people cheat on everything right now.

37:07I think they're focused on sales calls and he's attracted a lot of controversy and Rise is messing with him a little bit by just using AI to spam Roy with an almost inconceivable number of applicants. Somehow, I don't think it's going to affect Roy. I think he's going to be just fine. Roy's built different. He's hiring a videographer in San Francisco with 300 to 500K. Absolutely massive numbers for a vlogger. So if you want to vlog and make half a mil a year, head over to Cluey. Head over to Cluey. Good luck to them. Anyway, we should have our first guest joining right now, or set of guests, actually.

37:49I think we'll have two voices on the show. They're both anonymous. They've been involved in the latest report from the Stanford Review about spying on campus. You're not going to see their face. You're not going to hear their names, but you're going to hear us ask some questions about the report and try to get to the bottom of what's going on on campus and dig a little bit deeper into the viral Stanford Review article that went up most recently, just a few days ago. Anyway, welcome to the show. Can you hear us? Can we hear you? Yes. Thanks for having us. Thanks for having us. Great to be here.

38:25Great to have you. Thanks for joining. Can you start by, I don't know how you want to characterize your involvement in the story, or I know you want to remain anonymous, but can you just give us the high level on how the report came together and what the key findings were? Yeah, let's do it. So the Stanford Review spent over one year investigating Chinese academic espionage at Stanford. What we did is we talked to Stanford students, we talked to Stanford faculty, we talked to Stanford China experts, and we talked to congresspeople about this. And at the end of our investigation, we compiled a bunch of anonymous reports for people working in AI labs, student researchers and faculty.

39:01And the overwhelming conclusion was that there is essentially widespread intelligence gathering at the behest of the CCP at Stanford. What does that actually look like? I mean, no offense to Stanford folks, but like it's undergrad. It's not exactly like a proprietary AI lab or like nuclear weapons program. Like what are they trying to steal? Got you. So I'm going to pass you over to my colleague and she's going to talk a little bit about what her experience with espionage looked like sure yeah so a lot of um what espionage looks like you know on an undergrad level is um chinese students and researchers trying to basically collect any and all information about stanford that they possibly can so you know stanford is an open research institution so a lot of the things that are being reported back to the CCP aren't exactly secrets.

39:59And, you know, obviously, you know, it's not as bad as stealing defense secrets. But the issues that it's happening at such a massive scale, because they, the CCP essentially has all of their students reporting back to the CCP with any and all information that they have. Right. So once one Chinese international student characterized it like this, many Chinese students have handlers. The CCP wants to know everything that's going on at Stanford. This is a very normal thing. They just relay the information they have. Now, with regards to sensitive research, we're not just talking about the undergraduate level.

40:37We're talking about the graduate level. And these are some of the best AI labs and best robotics labs in the United States of America and the world at large. What they want is not just published papers, because you can imagine you can't replicate papers just via post papers. They want the methodology sent back. They sent back communication channels. They send back people that are involved with the research and other Chinese internationals that are working on it so that they can replicate this research at China. How would you rate the university's reaction to the piece? There's obviously been a bunch of conversation and chatter online, but I imagine you guys have been having conversations as well.

41:15yeah so we actually just published today um larry diamond matthew turpin um wanted to publish sort of a response article um so that was also published in the stanford review and uh they they came out in support uh of our article um which you know was good and then stanford also came out with a statement saying you know they take this very seriously and it is important to them. You know, we are sort of under the impression that there's not much that's going to be done about it because on some level, everyone knows that this is happening and nothing is really being done about it on a university level.

41:58They really want to kind of stay out of things. Right. I mean, Matt Pottinger, one of the guys that commented is the former U.S. Deputy National Security Advisor, Matthew Turpin, worked as the China Senior Advisor to the National Security Council. These guys have seen this happening for years on years. The issue is that Stanford is an open knowledge research institution. That means there's no secrets. And that means it's very hard to prosecute people for sending back sensitive research information to China. So really what they get people on is just their visa. They're working for the PLA, but they haven't told the authorities about it.

42:33So what needs to happen is we need to have people that are sending back this information registered to foreign agents and set up new laws and new legal guidelines that prevent people from being able to send public information at research institutions back to foreign governments. Right, yeah, and the other thing too is, you know, we also kind of need to recognize that the students are actually victims in this situation. I mean, the CCP is exploiting them. They are essentially crowdsourcing, you know, espionage at massive scale at, you know, universities all around the country. And they are, there's this thing called transnational repression.

43:13And basically, you know, if they don't want to report back to the CCP, if they have any reluctancies, or if they don't comply, their families can be threatened, their livelihoods, their scholarships. So, you know, it's really a sad situation for them as well. Yeah, I mean, totally. Matthew Turpin, who worked with the NSA, even said at Stanford, there's been Chinese international students who had their parents brought to the police station because they refused to turn over Stanford's research information. So if you don't comply, you will face penalties from the Chinese Communist Party and your parents and your family back at home in China may be put in harm's way.

43:51And we've seen that happen. Makes sense. Yeah, it's incredibly challenging. You could be here, your parents could be here, and yet if you have even one family member at home, a grandparent, a great grandparent, there's some amount of risk. I don't know if you saw this post by Kim Mike Cutler, who I think is at Partnered Initialized. She said, this report has pretty much all anonymous sourcing, unfortunately, and so it does not seem like a good platform upon which to argue that many, most or all Chinese nationals or first gen Chinese students are spies. How do you interpret that kind of criticism of the piece?

44:32Right. Let's go through this point by point. So first of all, when you look at how the CCP operates, they have 2017 national security law. Under Article 7, all Chinese citizens must comply with national intelligence and provide information and keep secrets when they're asked. So straight out of the gates, all citizens must comply at the face of penalty, working with security services. So anybody who's asked at Stanford to do something for the CCP must do it, or they will face legal penalties in China. The second point, we have talked to Chinese international students, they have families at home, for the same reason that there's transnational oppression, they cannot go on the record and say these things with their name out there because their families will be taken away their families have brought to police stations it's already happened with research with regards to chinese china experts we've had matthew turpin on the nsa we've had matt pottinger and the sn snsa and then we've had larry diamond all come out in support of our article they've said this is a decade long decades long relentless um interference from stanford to misappropriate our sensitive research technologies all the experts have agreed with us but at stanford we've seen a culture of silence and fear talking about this issue because it's characterized as racism.

45:45So 166 Stanford professors wrote an article to the Department of Justice saying that we need to shut down the China Initiative, which was an initiative that sought out Chinese spies because it's racist. But what that fails to recognize is if you don't shut this down, you have Chinese students who are the real victims here. So it's China who is profiling their citizens and victimizing them. It's not the USA. This is not racial profiling because we're just looking towards Chinese nationals who are being repressed by their own government. That makes sense. Putting aside the geopolitical issue of China specifically, how are you thinking about not suffering from a lack of brain drain?

46:33I feel like the majority of tech companies, like the tech industry might collapse if we didn't have, you know, a fluid transition from like, you know, there's Indian CEOs all over Silicon Valley. There are plenty of allied countries that send talented individuals, you know, even going back through history, we've had Operation Paperclip. We have historically successfully integrated members from even rival nations. And that feels like something that would be a risk of losing um if we go too far here is that is that a real risk or how do you think about um confronting that yeah i mean i think you know a lot of people have sort of interpreted our article you know and have said like ban all chinese nationals and from attending elite universities or you know coming to school um in the u.s um and we are we do not want that i mean you know i I grew up going to a Chinese immersion school.

47:37I think it's a beautiful culture. The people are amazing. And they are huge assets to Silicon Valley and Stanford. We just want to see some change in the way that these students are able to live and study abroad. Because right now, it's as if they have no rights, because they have such a hard time coming forward. So we need to figure out a policy system that can give these students a greater awareness of the freedoms that they have. And we need to be able to defend them. Right. A couple of solutions to keep Chinese students in the United States and improve security. I mean, we should remove WeChat from the United States.

48:20That's a main method of communication with the Communist Party. It's an unsecure platform that's monitored. Secondly, we should consider AI visas for bringing Chinese people's families out of China and to the United States. because that removes a huge risk of transnational oppression. So not just getting the best Chinese researchers, but also getting their families and coming to the United States. And that's what we saw with the Soviet Union. I mean, with the Soviet Union, whole people's families came with them. And then the third thing that you have to think about is China requires a lot of these people to come back.

48:48The USA should be doing the same thing. If you want to come study and work on the most sensitive research technologies that are key to winning the AI race, you should be forced to stay in the United States for 10 years or more Because what we really don't want to happen is for these people to replicate these ecosystems in China. Yeah. Who have you guys seen the most support from on campus? And you can keep it high level to protect the identities if people don't necessarily want to be public yet. but I imagine there's quite a few people on campus that have, you know, massive concerns around, you know, the national security issues that you guys are presenting and are, you know, hugely in favor of reform.

49:32Yeah, I think we definitely, you know, the vast majority of the response has been off campus. I feel like there's been a surprisingly little amount of response on campus, but I think the best reactions have definitely been from the hoover institution they have been looking into this kind of thing for a long time um and are working on you know policies to to deal with these things um but you know a lot of people that we interviewed still wanted to remain anonymous so it's really tricky because if you want to work with china on some level you can't be you know, consistently speaking out against China.

50:15So I think that's sort of the difficult balance. Is there an argument here for just kind of going completely open source with our research? I mean, famously, like the Manhattan Project had a ton of spies on it. We still kind of won the Cold War. It feels like a sloppy and potentially crazy solution. But is there a world where the only thing that matters in a geopolitical technology race is just pace of play and the actual secrets don't matter that much? Yeah, so you definitely want to play more offense and defense when you're talking about the technology race. We're not going to win this by stifling competition in America.

50:54We were going to win this through the way we've always won it by having an open free market system that encourages competition. You don't want to ruin the risk of China using vast government intervention and vast government subsidies to essentially take the best research out of America and distribute it writ large at China by using vast state apparatus. We've seen this happen with electric vehicles. We've seen this happen with most American technologies. China misappropriates it. I mean, 33 % of Chinese GDP is directed towards subsidies alone. The whole Chinese economy is geared towards leveling up and distributing advanced technologies.

51:30The issue with the United States. And it's a great thing is that the government will not subsidize these technologies to the same extent and encourage their diffusion to the same extent. So the real issue is that China may get these technologies and then use these subsidies to create a mass market in China and gain an advantage. So to a certain extent, there has to be sensible research policy that protects our core interests. After seeing all this, do you think that American tech companies are generally too naive when it comes to espionage? I think the major players in national security, defense tech, things like that have historically been very aware of the espionage risk.

52:09But do you believe that many sort of more American tech companies, specifically in the Bay, should be paying a lot more attention to security? 100%. The Foreign Interference Task Force was disbanded. The China Initiative was disbanded. Two key efforts to stop Chinese research and corporate espionage. We have very few lines of defenses right now. Companies need to be vigilant because we've seen this happen again and again. I would say that they know it's happening and they've known for a very long time. Even at Stanford, the people we talk to will all say, yeah, this is not new to me. Duh, we knew about this for so long.

52:47This is obvious. And they're surprised that we've even published this article to some people because it's so apparent. So what really needs to happen is we need the government to come through with sensible research policies that also encourage innovation and bay area companies particularly startups working in this area need to be more vigilant about this that makes no sense great well thank you guys for coming on and sharing it's it's uh extremely insightful and a bit uh it's scary it's scary i'm glad you're on the case it seems like you're researching this i'm optimistic uh we need people like you guys to you know keep banging the drum and and uh forever going to get changed so yeah Come back on when you guys have more news.

53:26And good luck with the follow-up reporting. Great. Thank you so much. We'll talk to you soon. Thanks so much. Bye. Next up, we have Morgan Housel coming in. A little bit changing. Switching gears. Well, I mean, geopolitics are in his wheelhouse. He was falsely accused of claiming that the U.S. would fall. Remember? Oh, yeah. So we got to talk to him about that. Addresses allegations of being anti-America. So funny. But we're excited to have Morgan on the show. He was great last time, and I'm sure we'll have a fantastic conversation. I'm excited to ask him about his new book, Art of Spending Money.

54:01Art of Spending Money. Well, welcome to the show, Morgan. How are you doing? Hey, guys. Good to see you. Great to have you on. Welcome back. It's been too long. Yeah. I mean, I want to start with the accusations that you claim that America was in decline. I saw a viral post. I didn't look at any of the replies, so I don't really know. What was the story there? It's weird because everyone knows that things get exaggerated on the internet, if not just made up on the internet. When it happens to you, it's pretty strange. So I went on the Diary of the CEO podcast. It aired, I think, a week or two ago.

54:35And we talked about tariffs. And I mentioned something about the fall in manufacturing jobs over the last 80 years, something to that extent. And I had so many different asterisks of like, oh, well, part is technology. Some of it is offshore, et cetera, et cetera. And that got spun into Morgan Housel is predicting the collapse of America. And just one of those just like, what? How does that even happen? And so I called the guy out. He deleted the post. But it's amazing to see when it happens to you personally because then you start questioning every like, how many times have I read a headline that said so-and-so predicts the collapse of America, but they didn't or some version of that.

55:15So I'm not predicting the collapse of America. I'm quite optimistic. Such a challenge right now. there's so much long form content that's created and we're also in this era of like clipping, right? The internet likes long form content, but it really likes, you know, these short segments and sort of pulling out interesting moments. And we've had some issues with that already where, you know, we had the CEO of Perplexity on and John asked him a hypothetical question about ads and TechCrunch ran with this article saying that, you know, perplexity planned to, you know, just jam a bunch of ads in their browser or something like that.

55:53And anyways, context is very important. Were you at the Berkshire annual meeting? No, this is the first one in four years I didn't go to. Awesome one to miss, right? Yeah. How was it? What made you decide not to go this year? Just family stuff, busy? I've got lots of other stuff going on. What's interesting is that I think I've been seven times and the last four times I went, I didn't even go inside to the meeting. It's turned out to just be a place where there's so many like-minded people who go there. So several years ago, me and Brent Bishore and Patrick O'Shaughnessy and Shane Parrish all rented a house and hung out.

56:33And it was an awesome weekend, even though we didn't even go to the meeting. So it's just been a place where there's so many collective people who have the same thoughts, same priorities to go and meet. that's great um well i wanted to i mean i reached out to have you on earlier this week because uh obviously the news of um of uh buffett you know stepping back at the end of the year to kind of get your reaction to it uh i it's one of those things it's interesting dynamic where uh it feels like the most important sort of knowledge and wisdom in the world is sometimes become so widespread that people, uh, sometimes people don't even pay enough attention to it.

57:15Even, even, even zero to one is one of those things that has like, it's, it's, it's one of the most popular, uh, business books in history yet. Um, people kind of underestimate the value of it because it's just become, um, you know, the ideas and it had become shared so broadly. So I wanted to have you on, you know, specifically to talk about kind of the ideas that, that you've, um, you know, most gravitated to from Buffett yourself. Some of the ones I'm sure are very widely, you know, understood and talked about. And then I'm sure others that are a little bit more kind of under the radar. There's just so there's so many good bits.

57:54So yeah, I mean, the first is that to put his retirement into context, Buffett made his first investment before Pearl Harbor. And he started professionally managing money as a professional hedge fund manager when Harry Truman was president. So just to put like the context of how long he's been doing this is absurd. And then, so there's two parts of that. One is like, yes, let the guy retire. He's 94. He's been going nonstop since he was 11. Like, come on. And then the other side of that is that's why he's successful. Yes. His annual returns are good. They were very good back in the fifties, sixties, and seventies.

58:26He hasn't really outperformed in any meaningful sense in a quarter century, not a criticism because he has almost a trillion dollars in assets now. It's almost impossible. It's almost impossible to outperform when you're that big. But the point that is so easy to overlook is that yes, he's been a good investor, but he's been a good investor for 80 years. And that is literally 99 % of why he's been successful. So I made this point in my book, the psychology of money that if Buffett retired at age 60, you would have never heard of the guy. He never would have been a household name. He would have retired with like a hundred million bucks.

58:57Awesome. He buys a yacht and a house in Miami and lives happily ever after, but no one would have ever heard of him. The whole reason he's successful is because he's been going for so long. And I forget who mentioned this. I forget where I read this, that Apple is Steve Jobs with a thousand lives. That's what Apple is. Like Steve Jobs built the company and then, and then died, but Apple can live on because he built what it is. And I think that's what Berkshire has always been. And Buffett has talked about this, that he wanted to build a company that would way outlast him. And the irony is like, he, he still last, he still ran the company for half a century or more.

59:29And so he lasted a long time, but it's going to keep going so far after he's gone, which is rare and unique. And I also think of all the lessons that. people like us try to learn from him. The wrong lessons that you can learn from Buffett is how to pick stocks. I think that's probably the wrong lesson to learn because a lot of what he did in his heyday in the 60s and 70s just would not work today. It's a very different world, different markets, faster information. It's just not transferable. But the lesson of he's successful because he stuck around for so long, even when he was so preposterously financially independent by hundreds of orders of magnitude, he kept going.

1:00:10And like, that's something that ordinary people can stick around on, like stick around long enough that you're going to let compounding actually work in your favor. I think that's probably the most pertinent takeaway for people. There's also so many entrepreneurs too, who, when they talk about, and I understand this, it's not a criticism because being a founder of a startup is so ridiculously hard and stressful. But for so many of them, the goal is we're going to build this company. We're going to scale this company and then we're going to sell it. And I get that. I don't look down upon that. it.

1:00:38But the huge, massive results are for people who are like, I'm going to do this and I'm going to do it as long as humanly possible that I can. Yeah, I have a bunch of follow-ups there. First up, Greg Abel, is he underrated? It's interesting because he's coming in in his 60s. He looks like a guy who could be retiring next year. But the culture of Buffett, you have to imagine that he's thinking, yeah, I got at least three more decades in here. I'm just getting started. So what's your take on Greg Abel? I don't know if you've actually dug into his career at all, but I'd love to know kind of how you think the culture that Buffett created kind of lives on.

1:01:19Yeah, he's 60, so he can run the company for 30 years and then run for president after that. That's how this works. That's the standard path. I think most people outside of Berkshire don't know that much about Greg Abel. he hasn't he's not like Buffett has very intentionally I think been on kind of a media tour for the last 25 years goes on CNBC all the time of course the annual meetings and the letters like people know a lot about him Greg Abel people don't know that much about him other than that he's been at Berkshire for 25 years and Buffett picked him as the obvious successor what is known is that he's not a stock picker and there's going to be people who who get into a lot of trouble over the next five or ten years who are like looking for Greg Abel stock picks and they're they're not going to find them anywhere.

1:01:59He's an operator. He's a damn good operator. Um, and that's important because Berkshire 30 or 40 years ago, its market cap was like half stocks. It like, that's like its portfolio was just public stocks that own Coca-Cola and Procter and Gamble and stuff today is like less than 20%. So the majority of Berkshire are wholly owned operating businesses and that's Greg's bread and butter. It just like operating those. I think it's, it's probably a very similar transition from Steve jobs to Tim cook. Steve jobs was this like, genius magician and Tim Cook was just a stone cold operator. And that's, I think that's what we're going through with Berkshire.

1:02:35You're going from like the, the wizard stock picker magician to the stone cold operator and worked out great for Apple. It's just, you have to keep your expectations in check both because Berkshire size and because of what Greg, Greg Abel strengths are. It's not going to be, and Berkshire hasn't been this in 20 or 30 years, the place that's going to have like massive outperformance year after year. Yeah. I wonder about how much he needs to be a stock picker because I totally understand the operator lens. But when you're looking at that $300 billion cash pile, it's like, what are you going to buy?

1:03:10A big oil company or something like you can do so much with that. So I'm sure that will be a challenge. I wanted to talk about the legacy of Warren Buffett a little bit more. There was a retrospective in the Wall Street Journal that I kind of took issue with, but I wanted to get your take. The author is talking about Buffett's remarkable recall. He estimates that Buffett read more than 100 ,000 financial statements. He famously reads books and seven newspapers and just reads constantly. And the author says, his unparalleled exposure to financial information combined with his prodigious memory made Buffett into a human form of artificial intelligence.

1:03:48He could answer almost any query out of his own internal database. That has given him an unparalleled ability to identify the kernel of significance in any new bit of information and a durable advantage over other investors. Now, and this is the controversial part in my mind, says now that AI is universally available, a person with Buffett's massive command of data won't even have an advantage in the future. And I just don't know if that's true, but I wanna know, do you think AI changes the landscape in a way that the Buffett strategy just doesn't work anymore? I think part of the problem with Buffett in the last 20 or 30 years is that his folksy grandpa demeanor hid the fact that he is off the charts intelligent.

1:04:31And most of the time, when you think of like an Einstein kind of genius, like they have a certain look and they talk a certain way that, and Buffett was not, that's what made him so popular is you felt you could relate to him. But anyone who spent a lot of time with him, when he's not talking for the camera or writing in his folksy way will tell you he is he is in a different universe of intelligence, particularly for money. And the reason he's successful is that since he's been 11 years old, he has spent 24 hours a day, seven days a week thinking about stocks and nothing else. And so it's less about the data that he knows, even though people will tell you those stories, that he can recall specific figures from a balance sheet from an annual report he read 20 years ago.

1:05:10So there is like a just insane memory recall, but it's the pattern matching recognition in there. And then this other element that is way more important, which is his reputation. That's something that AI can't do. And so a lot of Buffett's biggest investments, certainly the most important ones, came from the fact that he had such a good reputation that he could walk into a boardroom and just say, hey, I'd like to buy 10 % of your company. And they were like, yes, name us your terms. We'd love to partner with Warren Buffett. And so that's less about intelligence or data and more about the reputation that he had.

1:05:42And I think if you, I think if Buffett were a jerk or a raider and had a reputation of stripping companies, he would not have been 5 % as successful as he was. All the big deals came from people wanting to partner with him. Particularly, like think about 2008, the financial crisis. Every bank called him up and they were like, Warren, name your terms. Like, just tell your terms. And he did it at Goldman Sachs and GE and Bank of America where he got these crazy deals that nobody else could have because of his reputation. And so I think there is a - So it wasn't just that he had the liquidity, you're saying?

1:06:18Right. It was just like - Yeah, there's a lot of liquidity in the world. There's a lot of people who can write a big check, not who can do it with that kind of reputation. So I think there is a little bit of truth to that idea that the skill that he had of reading annual reports was so much more valuable in the 1970s than it is today because everyone's reading the same reports. There's bots that can scam them just instantly. But that's not to say that anyone can be a Warren Buffett now. I think that's certainly a stretch. What do you think of this narrative that the past decade of Berkshire is really just the story of Apple and, again, kind of a stock pick there?

1:06:58They're at such a huge scale. Do you think there's a cultural shift towards more technology? Of course, when Apple was already the most valuable company in the world. and it still outperformed the market and drove fantastic returns. It seemed like a big shift at the time. At the same time, it panned out very well. Do you think that that represents like a significant change in the culture? I think what's interesting is that for having a reputation for decades of, I don't do tech, I don't do tech, I'm a grandpa dinosaur, I don't do tech. Literally in dollar terms, Warren Buffett is the most successful tech investor ever.

1:07:34He made a hundred billion dollar profit on Apple. No one's even come close to that. And so is that like a change or an evolution? It is an evolution, but it's easy to assume and it's wrong to assume that Buffett's been the same investor for 80 years. I think part of why he's been so successful is that every five or 10 years, he completely updated his operating system to have a different style, a different influence and to really adapt what was going on. And I think that there's only a handful of investors who are like that, who have made money in different eras. Uh, there's a lot of investors who were very well suited for one era.

1:08:10They could make a lot of money for 10 years, but they couldn't keep it going because nine times out of 10, it's because they didn't update their, their thinking. They were stuck on this world that didn't exist anymore. And so there's a very long history of Buffett doing that in every 10 years, doing things that he would not have 10 years before that. Yeah. Are you familiar with that famous napkin diagram that Walt Disney drew showing how the parks relate to the film production, relate to the merchandise? Have you seen that chart before? No, but it sounds awesome. Okay. So Walt Disney drew this big chart mapping the entire ecosystem of Disney and how everything works together.

1:08:51And it's cited as this example of like, oh, like, you know, build this business where everything feeds into the other piece. And, you know, what is your Disney map? but I was digging into it and I realized that Disney made that chart 10 years before he died like he had already been building just in film for a decade and then 20 years to build the parks Walt Disney born in 1903 I believe the park didn't open until 55 and he died like shortly after and so it was more like a reflection on what he had built as opposed to it was a map that was that was drawn after he had explored the territory. And so I'm wondering your take on this idea that like the idea of focus and the idea of empire building being something that can be charted beforehand versus has to be kind of naturally discovered and then can be potentially mapped after the fact.

1:09:52I think there's several historical examples of business, of like very successful business people who were product geniuses and terrible at business. Walt Disney was one of them. He was the product genius of all time. He was a terrible businessman. Henry Ford was another, Henry Ford is the most successful mechanical engineer in history. And he was at best a mediocre businessman. There's so many of those like that. I would even say Steve Jobs might fall into that category of technological genius, design genius, very, at best, mediocre business person, which is why you needed someone like Tim Cook.

1:10:29And I think there's been quite a bit of that. But when the product is so good, you can take it in so many different directions. Disney is an interesting example of so many of the film IP that they had that was sitting on the shelf for decades. And when it was made, it was like, oh, people will watch this in the movie theater. That was the only medium that they could watch it on. And then the video cassette tape came along and they're like, oh, we could sell all of this. And the DVD came along. They're like, oh, there's another avenue to sell it. And then streaming came along and it was like, sky's the limit.

1:10:59So it was like the, the IP was so good that you didn't need a genius businessman to, uh, to, to come up with a distribution strategy. Like it was so good that it just kind of ran itself. Ford was, was similar. Like the cars were so superior to anything else that even if Henry Ford was making blunder after blunder, the company kept going just fine. And so I think, I think a lot of times when you have like a very long running business, it's less because they made phenomenal business decisions and more just because the product that they made was so phenomenal that they could keep it going. If an idiot ran Apple, it would still sell a zillion iPhones every single year.

1:11:36But there's only a handful of companies that are like that. I want to talk about the book out October 7th. Do you have a log line or like what is the one line pitch? And then how can you unpack the title is the title is the title is the pitch, but can you, can you break it down into a little bit more detail? A little bit detail. The book is not called the science of spending money because I don't think that exists. There's no way of saying, here's how you should do it. That's going to work for me and work for you. Even if we're similar people, it's just like spending is very individualistic. So I call it the art of spending money because art is different from person to person.

1:12:10It's, it's, it's subjective. It's often contradictory. And so rather than telling you what to do, I don't want to lecture anyone, but the book is a look at envy and social aspiration and keeping up with the Joneses and becoming fulfilled and getting attention and watching other people. It's kind of like the psychology of spending money, which has always been like so much of what I write is just trying to figure out my own life. So I looked at instances where I was clearly, it's hard to admit, but like when I was clearly envious and when I was clearly trying to get people's attention. And just when you dig a couple layers below that, like why, like whose attention am I trying to get?

1:12:48Is that person even paying attention to me? Do they even care? Was giving that attention, like taking away from other parts of my life? There are so many different layers to dig through. And I think those things are universal. Even if what you're spending money on is not a science, the psychology of it tends to be pretty universal. So it's just a look at that. That was more than one sentence. I have to hone that a little bit sharper, but that's it. That's great. I mean, is there a key anecdote or story or even just like a case study that you think is the most tractable for people to grab onto?

1:13:19I think about, you know, all the emotion that goes into the identity tied to the car that you drive or the house that you buy. But what case studies are you pulling from to kind of ground the lessons in something that anyone can understand who's reading the book? I don't know if this is the best case study, but it was one that I thought was so interesting. I read the biography of Harvey Firestone from Firestone Tire. He was the tire magnet like 120 years ago, whenever he lived. And he has this part in his biography where he says every single successful person that he knows, including himself, when they became rich, they bought a giant mansion and every single one of them to a T hated it.

1:13:57It was a pain in the ass to run. It was, it was, you know, when, when, when, when your house is 17 bedrooms, it's just more roof to leak more radiators to break. And he was like, everyone hates it, but all of them do it. Every single one of them does it. And even when they hate the house, they never sell it and get a small house. And he had this line that I loved. And he said, there is no going back except as a broken man. It's like, once you inflate your lifestyle, you cannot deflate. Even if you hate the inflated lifestyle, you cannot go backwards because it becomes so synonymous with your identity.

1:14:28It's like the size of your house is who you are. I thought that was pretty interesting. And he was like the equivalent of a billionaire living in a mansion. It's not a very relatable example. But I think there's so much of that, that if we spend money on something and it doesn't make us happy, it's very hard to rewind. So be like, be really careful when you're inflating your lifestyle because it's very easy to go forward. It's extremely hard to go back. How much time have you spent or are you putting any attention in the book towards the way that San Francisco and the tech industry spend money?

1:14:58I've always been fascinated by it, given how much just this dichotomy between extreme wealth and this, in many ways, desire. One glass box, please. Yeah. Yeah. Or this desire to, you know, be and it's good in many ways. Right. If tech, if imagine that the negative attention that the technology industry would have gotten if every, you know, Series C founder in San Francisco was like driving a Lamborghini. Right. It would have been fun, but it would have been. Well, I think there's part of that, that they can't spend a lot of money because it's not liquid. Like they're rich on paper, but there's not a lot of liquid wealth in San Francisco.

1:15:40I mean, there is, but it's all relative. But even that is a sense of social signaling. Like if people know that you're rich and you are going out of your way to live an austere lifestyle, that might be because it'll make you happier. It might also be because that's the signal that you're trying to send. That's like those are your peacock feathers. others. Buffett's the best example of this. I mean, you have to spend some time on Buffett in the book of just intentionally. Yeah. But then that got kind of twisted because Mark Zuckerberg was famously driving like a Corolla for a while, but then Sam Bankman-Fried at FTX kind of used that as window dressing to be like, oh, I'm the altruistic billionaire.

1:16:18I just drive a beater car. Right. And of course he had like a mansion in the Bahamas and a bunch of other stuff that he wasn't pointing the camera at. And Mark Zuckerberg's interesting because he famously drove the Acrea for whatever, for, for many years. And now he just bought a half billion dollar yacht. So it all, it all, it all comes around eventually. Yeah. He's also evolved his taste in cars a lot where he could buy a supercar, but I believe the most recent car he bought was a Cadillac CTSV Blackwing, which is like a hundred thousand dollar sports car, but it's not a half a million dollar sports car, but it's an American, you know, muscle car.

1:16:50It's like very fun and very different. So he's clearly finding his own path in expressing himself in a more unique way than just like one rich guy car, please. And so I think he's probably happier and like carved out a unique niche. And I feel like that's often a better place to land is not just don't just buy the most expensive thing. It's okay to buy the expensive thing if it's craft and it's interesting to you and it has a story and it has relevancy or performance. But it needs to like speak to you in some unique way that actually, you know, improves your life. I don't know. What's your most irrational uh purchase i know you've talked in psychology of money that you you paid cash for your house or you don't have a mortgage i believe but uh what else is is potentially irrational i don't know if this is irrational that's what i think about a lot i grew up skiing uh as as a competitive ski racer and when i was you know 10 12 13 years old i always felt like all my friends had better gear than i did they had the nicer skis the newer jacket and whatnot and back then it bothered me when I was 13.

1:17:52I was like, I was so envious of them. And so now, and obviously I got over that, but now that my son, he's nine. Now we, we ski a lot together. And I think because of the scars that I had for my friends, having nicer stuff, I made this vow a couple of years ago. I was like, I'm going to buy my son the best skier every year. He's going to get the best skis, the best boots, the best jacket. And I'm doing that to kind of like fill this hole that I had when I was, when I was a young kid. And the irony about it is he could care less. He does not care whatsoever what gear he has. And so that's interesting too.

1:18:23Like, I think there's always a story behind spending for me. It's like, he doesn't care because he has no, he doesn't care because he's so all he knows is the best, right? Is that, isn't it? I feel like, I feel like I, like if, if I buy him a new pair of skis, I expect him to be like, Oh wow, let me look at these. They're so cool. Cause that's what I would have done. And I get them and he's like, this doesn't, doesn't matter whatsoever. No, I totally resonate with that though. Cause I remember growing up skiing snowboarding i would i would always have a snowboard that was like three seasons old lightly used and it was like the practical decision it was fifty dollars it got you down the mountain just pretty much as well as anything else but then i'd be watching the burton you know videos and it's like you know looking at these and i'd be just running the numbers i'm like i would have to ref 400 soccer games to get you know that board with those bindings it's like i'm gonna ref like 10 games this weekend but i'm not gonna get there i was skiing at some point i got too tall for regular skis and so they needed to they're like you have to go to custom skis if you want to be really in this and i was like i'm switching to scuba diving like you know whales are big enough for the ocean i'll be big enough for the ocean it's fine uh i do have one last question do you have one i have one last too but um there you go switching i hate to switch uh back but on the art of spending money what is the uh what do the real uh berkshire enthusiasts think that uh or what is berkshire signaled around how they're going to spend 300 billion dollars or whatever their their current cash pile is because i feel like the general market sentiment is that you know berkshire is preparing for the next you know uh next time great companies go on sale due to some crisis.

1:20:10But I'm curious if you have any sort of insight there. I think the list of companies that you could acquire for$100 billion is not a very long list, but there is a list. It's not inconceivable that Berkshire can make$100 billion acquisition. I've often thought that Bloomberg, the media company, that would be a perfect Bloomberg fit. It's probably worth about$100 billion. That'd be perfect for Berkshire. And then Berkshire is so big, it's a trillion dollar market cap, they could probably repurchase a hundred billion dollars of stock. It might take three or four years, but they could do it. There's enough liquidity there.

1:20:43So you could spend 200 billion. Like that's, that's pretty conceivable. The big question for me is once Buffett is gone and passed away and his, his stake in Berkshire is more dispersed, then you could, there's room for an activist investor to come in right now. You can't have an activist at Berkshire because Buffett owns too much. He would just tell him to go away and And then it's over. Once Buffett's majority voting stake or like big voting stake is gone, then it's interesting. Will Bill Ackman or one of those guys come in and say, we want you to do a$200 billion special dividend. We want you to spin off this and that and that.

1:21:19I hope that doesn't happen, but you can very easily see it happen, particularly if people aren't going to give Greg a five-year leash to prove his way. If the results aren't there in two or three years, you'll probably have people knock at the door. And so having that pressure, I hope he doesn't have pressure to deploy that$300 billion of cash, knowing that if he doesn't, someone's going to come knock at the door and threaten his job or tell him to spend off this or that. So it'll be interesting. You said he's not a stock picker either. So he's not exactly just going to go out and be like, okay, I'm going to start deploying this and that way.

1:21:55It feels like it has to be more significant. Yeah. Yeah. I have another question, but we'll have to do it next time. This was great joining us. Leaving us all hanging, John. I know it's just going to turn into a whole conversation and we can do a whole another 30 minutes. So why don't we just do another 30 minutes in a couple of weeks? Let's do it. I'll see you then. Morgan, we love chatting with you. Thanks so much for coming on. Thanks for coming on. Have a great Friday. We'll talk to you soon. Bye. Here's the run of today's show. We have Sonia from Sequoia coming in the studio next. Sequoia just hosted the AI Ascent Conference with an absolutely stacked roster.

1:22:27are some of the greatest programmers, some of the greatest entrepreneurs in artificial intelligence, or really just in history, all coming together. We saw a fantastic jacket swap between Alfred Lin and Jensen Wong. And I'm excited to talk to her about trends in artificial intelligence and what is happening in both the early stage and mid-stage startup market. Obviously, as a venture capitalist, she is investing in a lot of interesting stuff today. So welcome to the stream, Sonia. So good to meet you. Thanks for having me. you know alfred was bummed his jacket was hermes and he was sad to lose it and i was like don't worry i think i think you got the better trade here oh wait so that was a permanent swap he gets to take it home permanent swap okay that's fantastic yeah it needs to be framed i mean it's iconic iconic yeah i mean i want to buy a leather jacket just so if i'm ever in the same room i have something to swap yeah we'll get you the dates don't worry fantastic uh yeah give me the rundown of ai ascent is this something that happens every year was this year special obviously it's a you a bigger trend than ever?

1:23:27Who were some of the interesting speakers? Give us the overview. Well, I'll give you the origin story. I'll take us back a bit. We invested in OpenAI back in 2021. So this was back when it was very much, you know, it was a few guys using the API, definitely no ChatGPT yet. But we just felt like they had invented magic. And we wanted to make sure that our portfolio companies would be kind of the first to be able to see that, play with it, and transform their own businesses. And so we set up a field trip actually for like 40 of our portfolio companies to go visit OpenAI back in May, 2022. And so that was like pre-ChatGPT moment and everyone loved that field trip.

1:24:04It was like people were playing with Dali for the very first time. You know, there was no public access yet. Starting to build with GPT-3. And so our founders loved it both for like the inspiration element of like, oh my gosh, we are in the belly of the beast of the thing that is building the building the magic but also like from a very tactical perspective like here's how we should be using this stuff to transform our businesses and so founders loved it they asked you know can we bring this back next year i honestly i'm not like a party thrower i like hate throwing events and so i was like oh man we have to do it again but everyone to do it so we um we did our first kind of non-open ai specific event um and more across the entire ecosystem the following year.

1:24:46That was right after the Chachi Petit moment. And so this is our third year throwing that event in a row. We've had amazing speakers. Sam Altman has spoken at every event. We've had Jensen twice. We've had... Even the audience is incredible. In our opening talk, we called out what we viewed as the biggest AI product innovations of the year. And we had Notebook LM, Deep Research, Sesame. I think some of the biggest innovations, those people were all just happened to be sitting in the audience. So it was just a lot of firepower in one room. That's great. What are the top kind of discussions that people are debating right now?

1:25:21I mean, from talking to people on the show, this idea of like the pre-training wall, needing to move into more RL focused techniques to kind of get the next level of breakthroughs. Is that the right question to even be asking? Were people debating it? And do you have a take on the idea of like this hitting a pre-training wall. Yeah, totally. So we had Noam Brown speak at last year's AI Sense and we had Dan Roberts this year. They're both on OpenAI's Strawberry team. And we had actually gotten a preview of this from Noam actually before he even joined OpenAI. And so he had done a lot of research historically in AI gameplay.

1:25:57And if you take the lesson from Go, for example, with AlphaGo, which I think was one of the seminal results in reinforcement learning um the top humans are like a 3 500 elo at go uh the best bots are like 3 000 before you give them access to inference time compute but if you if you actually let the model actually sit and think for a minute before it places its piece you can get that elo up to 5 500 points so like way better than superhuman um and so the key insight there is like um to get roughly the same order of performance once you've like it's diminishing marginal returns on on multiple vectors right but But once you've kind of like hit diminishing marginal returns on one vector pre-training, if you start scaling post-training from there, and I'm sorry, specifically inference time compute, you can get like 100 ,000x scale up in performance there.

1:26:44So we're just starting to climb that second curve. I think OpenAI deserves a ton of props for like seeing that and like investing decisively behind it. because I think if you talked a lot of research in the ecosystem, like one, a few years ago, it wasn't even obvious that the lessons from AlphaGo could even apply to the LLM world. And so, you know, like if you went to visit research labs a few years ago, it was like there's the RL group and there's the LLM group and they're not the same people. And it's like, it's very different. And I think Noam and various other people like really push forward that vision.

1:27:17And I think Sam invested heavily behind a reasoning infrastructure because the hard thing is like scaling up reasoning infrastructure is different. from scaling up pre-training infrastructure. And when I talk to my friends, a lot of them have joined certain labs like OpenAI, XAI, that have really invested ahead in reasoning infrastructure because it is such an important vector for scale. Yeah, I was always wondering about the AlphaGo pre-training kind of scaling law and wondering like, okay, we have all this major compute. What happens if we go back and train AlphaGo on 100 ,000 H100s? Like, are we going to get even better or really have we actually topped out?

1:27:53And it sounds like we basically did and we kind of learned that lesson. But is that a refutation of scale is all you need, kind of the bitter lesson? Or do you see it as just a continuation of that theme that we will need to continue scaling and it will just be new algorithmic paradigm on top of what we have and then scale that and then another one and then scale that and then another one and then scale that? Because it sounds like, you know, when I hear 100 ,000 X improvement in test time compute or inference time compute, that sounds like a lot of data centers. Yeah, it's a lot of data centers.

1:28:31I'm very much in AGI maxi, like pro, better lesson. And I think that this is, you know, just another vector that we're going to scale on. And it's not like pre-training is dead, right? It's like you kind of like, you know, from an economics one-on-one perspective, you go to like, where is the lowest marginal cost of the incremental unit of intelligence. And so right now, a lot of that is on reasoning, but I think it's going to break to other vectors as well. Are you seeing that in image and diffusion as well? It feels like the images in chat GPT is doing something different. It feels like they're layering a few different techniques together.

1:29:07I was playing with the text and trying to get it like the text is so good now, but I was trying to get like a snake to weave in and out of the text and it was kind of getting confused and I was like I feel like there's some layers going on here or something I'm trying to like kind of understand it and I'm just wondering like uh you know we might be past just like the big transformer paradigm of LLM and text responses in the reasoning era um are we are we evolving past the big diffusion model in image generation as well um so I'm not a researcher but I get to talk to a lot of smart researcher friends.

1:29:43My understanding is that it's a combination of a transformer diffusion architecture. And I think that most people don't believe that diffusion models will fully get us there, whereas transformers have a lot more juice in them. And so even if you look at, it's not just image, it's video, it's robotics. A lot of those have transformers as their backbone. But I do think there's so much that's happening in the harness around it, right? What's the for loop that you run the model in? What tools do you give it access too. We did a little pull at AI, a sense of like what, what innovation is going to drive the most amount of progress in the AI ecosystem in the next 12 months.

1:30:18And like biggest answer was MCP and tool use and forming an ecosystem around that. So I think like the models themselves get smarter, but they're also surrounded by a big ecosystem. On MCP, how are you viewing that as a position in the market? Is it just a standard? Is it just an API? Is it a framework or will there be companies that build around it? Are there going to be open source frameworks that then we find a red hat Linux of MCP and it winds up being a big company, even though it's mostly open source? How are you thinking about that from an investor's perspective? Yeah, I very much see it as a protocol and like something for the industry to standardize on.

1:31:00And so I think there are obviously some benefits of the crew to Anthropic from having steered that. But I think it being an open standard is really important. And that's why a lot of the other big model labs are standardizing behind it as well. And so I think it's a net positive for the ecosystem. There's a bunch of startups spinning up trying to make money off of it in some way. I don't know if I'm really bullish on any of them really having a shot at it. I mean, I think there's certain people that have, if you have an infrastructure advantage and for some reason, like Cloudflare, for example, is making a big play, then like maybe I buy that.

1:31:33But if you're a small startup trying to spin up a MCP shop, I just don't really see the right to win. Yeah, maybe the value accrues to like McKinsey coming in and saying like, hey, we're going to help you implement an MCP server for your existing Fortune 500 company or something. Was there any interesting conversations around benchmarks? And do you think they were talked about more or less than the last year? I would say this is an audience that very much knows that the benchmarks exist and doesn't really care too much about them. We ran a poll of, if you could only use one model for the rest of your life, what would it be?

1:32:09And OpenAI, by far, number one, more than 50 % share. Even though if you look at where they are on the model leaderboards, they're not there on LM Arena. And so I think there's a little bit of a, the benchmarks aren't really, they're saturated, they're not really the vibes test. I think people care a lot more about the Vibes test right now. Well, yeah, and just end user value, right? Yeah. Well, speaking of benchmark, do you have a take on the Manus investment that's kind of burning up the internet right now? Oh, man. Seems like an odd choice in 2025. I don't know if you looked at the deal. I'll give you a hot take, John.

1:32:45Apparently, the U.S. Treasury is examining a benchmark capital's ties to Manus. So feel free to pass on the question if you don't want to talk about it. I will say they've built really cool tech. I think the devil's in the details for what exactly, where's user data, et cetera. And I would imagine that they did their homework, but I don't know. Yeah. Switching gears, the big story this week around OpenAI, obviously after the event, was the new CEO specifically coming in to focus on applications. I'm curious, you know, if you could highlight any of the kind of conversations around value accrual.

1:33:26You know, Sam, I think, has been pretty explicit in the past that if you're building products with the assumption that the models are going to continue rapidly get better, you're probably in a good spot. If you're not, you know, maybe you're going to struggle or get made redundant at some point. But I'm curious what the general kind of vibe was around that. Yeah, I would say like, I mean, you know, the mimetic cry in the venture ecosystem right now is just like the value is in the application layer, the value is in the application layer. I actually have a great meme on this. We had the meme on like the values in the application layer and then we had Jensen in the audience and I just had a picture of Jensen on top of Scrooge McDuck just raking in all the dollars.

1:34:10But like we very much think values in the application layer and it depends how you want play it right um i think that like there's there's going to be a place at least in the near term uh for uh vertical agents applied to a very specific sector and like we've we've a bunch of those companies our portfolio sierra harvey open evidence um i think that what the foundation models have proven though um which was debatable a couple years ago is that they have every right to win the application layer um and so like it wouldn't have been obvious that you know a company building foundation models could figure out the application magic.

1:34:45But ChatGPT, to me, it's like a runaway freight train in terms of consumer adoption. Some of the metrics they've published, like 300 to 500 million weekly actives year to date. It's phenomenal user growth. One of the things that Sam shared at our conference in terms of how people are using ChatGPT was really interesting to me. If you're old, you're using it as a Google replacement. um so i'm old uh if you're in your 20s or 30s um people tend to be using it as like more of a life advisor life coach type thing and then if you're really young like the youths are using chat gpt as an operating system and i found that framework really interesting and you know especially in combination with like they're clearly building things around memory um around tool use around connecting to your your other applications it really does feel like you know if you're if you're a young person and like really, really connecting chat GPT and like mind melding with it.

1:35:40That use case seems really interesting to me. And so like, if you, if you, you know, if you think about Google as front door to the internet and$2 trillion market company, right. It feels to me that chat GPT is in many consumers' minds that front door to AI. And as what AI can do as the ceiling on that goes up, we just like each deepen our product usage. A couple years ago at the first AI sent, I posted this chart of the ratio of daily to monthly active users for ChatGPT and some of the other mainstream mobile apps. The punchline at the time was usage is terrible. It was like a 14 % down mile, if I remember.

1:36:22People kick the tires a little bit and then churn. We've been tracking externally from data science signals, just seeing those down mile ratios increase. it's pretty crazy like demo is now in line with reddit it's approaching google levels and so like if you think about reddit being like a super engaged like you know you're in there having like multiple conversations you see that same behavior both anecdotally and in the data for chat gpt and so to your question where will value accrue we think it's in the app layer i think a lot of the horizontal app layer opportunity will be will be won by foundation models like like openai like xai and then a lot of the companies we're backing are going after or very specific vertical opportunity.

1:37:02Are you particularly bullish on enterprise application as independent? I mean, I'm thinking about like Google very much one consumer search, but you know, in legal, there was LexisNexis and then Palantir helps, you know, the government search through data sets. And there's all these different enterprise use cases that Google wasn't able to go after. We were even talking about Armada, which is an enterprise like, you know, service built on Starlink. the Starlink team has dominated in consumer, but, uh, you know, the enterprise needs of certain enterprises is just too unique. And so there's actually a business to build there.

1:37:39And that seems like it maps with your strategy most recently, but what do you think? Totally. Totally. I think the, um, the enterprise war feels like it is like, you know, we're in the first inning still, um, it's not clear who's going to win yet. I think, you know, in our portfolio, like Glean, for example, has done a really amazing job as like a horizontal chat GPT like platform that kind of connects to all your enterprise data. But then it's like a question of like, how deep do you need to go? Like, you know, Harvey has a ton of legal case specific data or Sierra has a ton of like customer support specific data and workflows.

1:38:09And so I think there's, you know, the enterprise AI battlefield is very much a work in progress at this time. I think like the shape of workflows and problems is like so diverse in the enterprise that like my guess is a lot of these companies will be successful. Yeah. How are you thinking about humanoid robotics? I've talked to, there's a lot of high flying companies that feel like it's mostly renders at this point. There are some really amazing researchers working on stuff. I'm personally waiting not to see the demo of one humanoid walk around because we've seen those with Boston Dynamics for decades.

1:38:41I want to see the satellite photo of the data center that's getting built out to do a massive training run for an end-to-end robotics model. And I haven't seen that yet. Is that the right signal to be looking for for takeoff in humanoid robotics? Or should there be something else that I'm tracking as we go into this humanoid rollout, which it feels like it could be tomorrow or it could be two decades away? It doesn't feel like tomorrow to me. I was going to say the same thing. I'm being generous. There's a lot of CEOs out there who say this tomorrow. I can't share the details, but let's just say I think the humanoids are a lot closer than we may think they are.

1:39:17Okay. And it's like I thought this stuff was science fiction, and I've talked to a lot of smart people who have told me and shown me things that have made me realize like wow this is probably on the time like I would guess in like two or three years this stuff will be in tens of thousands of households at least maybe households yeah um and uh that actually echoes kind of like the timelines from that I think Sam sat on stage that you know um we had Jim Pham from NVIDIA on the on the show he like he's like you know there's the there's the digital Turing test of like you don't know if it's a human behind the uh the computer or not or not sorry a computer uh a human behind the screen when you're talking to it it's like the the physical turing test is you know when you leave your house and it's a mess in the morning and you come back and it's like it's all perfect and and all cleaned up um and like he also thinks we're we're about to get there um for him like a huge part of the breakthrough is is going to be synthetic data pipelines sure um and so like robotics unlike llm like you just don't have internet scale data to train on um but like one of the amazing things about what's happening in AI right now is just like with LLMs, but also more specifically with these generative world models, you can generate tens of thousands of variations of the same environment to simulate.

1:40:31And these robots can get better and break through that data barrier extremely quickly. And so I was personally like a robot bear for the longest time and talked to a bunch of very smart people in robotics over the last week. And I flipped. I feel like the human rights are coming. And in terms of capital intensiveness of like actually getting there in a few years, obviously it's very expensive to build a factory that produces robots. But do you think we're also going to see raises from humanoid robotics companies that are where, you know, a ton of the raise goes into NVIDIA GPUs to build a huge data center to train some massive model?

1:41:08Because yeah, I agree with you on the synthetic data. You could wind up with web scale, you know, trillions of tokens like we've seen with the GPT-4 training run. I think it's going to be extremely capital intensive. And it reminds me of autonomous vehicles five plus years ago, where like, you know, at the end of the day, like we have Waymos and we have Teslas driving around. Those companies had enormous economic engines to support the development. I think the same is happening with Humanoid right now. And so my guess would be it's like, it's not just NVIDIA GPUs, it's everything, right? because you're co-developing the hardware.

1:41:44You still have to collect a ton of data on actions. And so it's a ton of spend everywhere. Yeah. I mean, huge trends in AI broadly. It feels like all the metrics are up and to the right. At the same time, valuations are very high. What's your overall take on the venture market? Are we in a bubble right now? Can there ever be too much venture capital? All the key questions. Never, John. Never. There is too much venture capital already. Um, look, I'll say a lot of companies are raising on like what I'd call vibe revenue right now. And, and like, it's like, it's like, it's, you know, pilots being counted as revenue.

1:42:21It's like really, really terrible retention stuff. And so like, um, once you peel past that, I think there's like a cohort of companies that are like growing high quality revenue at the highest pace that we've ever seen. Um, and that includes, like, you know, I mentioned open AI before, but it's also companies like Glean and Harvey and Sierra and all these companies. And so to the extent like valuation is a function of like how much have you de-risked, how much product market fit do you have, like what is the growth rate of your business and what is the ultimate TAM potential? I think these AI companies are demonstrating just, you know, growth rates outside of what we've ever seen before.

1:42:57And then TAM potential, like because it is very much you're selling into, you know, if you're able to get outcomes based pricing, you're selling into a services replacement or a tools replacement. It's a TAM in the trillions, right? Okay, but. Yep. Sorry to interrupt you on that point. Something I'm curious about is, so yes, if you have an AI tool that can replace services spend, you can capture some, you can, whatever, ideally capture a lot of that market. But the thing I keep coming back to, and maybe this isn't the right way to think about it, but you're not simply competing with end humans that are delivering that services.

1:43:35A company will also be competing with other AI tools that have a similar cost structure. So does that not over time just drive the sort of dollar amount that you can capture just down to something that looks more like a software market? We have this debate all the time. um so i'm glad you bring it up um i think it really depends like if if what you're doing is like you know really low switching costs really low differentiation above what the models provide like yeah i think that margin is going to get competed down um and so i think that's why we've we've historically debated a lot of these gpt wrapper companies i think that you know if you're building something that's really hard to build um or that's integrating into a customer base customer base that wants to like choose an AI champion and move on with life.

1:44:24Which by the way happens, like we just did a bunch of references in the healthcare transcription market. Like you talk to these healthcare CIOs, they're like, I'm choosing one transcription vendor. I am not ripping that thing out for the rest of my life. And so I think there's, there's nuances to the stickiness of these things. Like I work with a company called Gong and they do, you know, they're a sales AI company. And the theory was all like always like transcription. Love it. We love gongs on the show. You baited us. It's a huge part of the brand. Oh my gosh, that's amazing. And they have such a quirky brand too.

1:44:58There's so many gongs around their office. That's amazing. But the theory was always transcription should commoditize. And I think that very much hasn't happened. And sales teams standardize on them. They standardize their processes on them. They train all their reps on it. And it has all your data. So I think the theory of how you build moats is different from rubber meets the road in in terms of like how these companies in practice do build the moats. But like, I think you are in a run like hell business because there's so, you know, there's, you know, we backed an AI DevOps company, like an AI troubleshooter.

1:45:30There's four other companies that are trying to do the same thing right now. And so like, we are kind of in like a run like hell segment of the market right now. How do you think about private equity stepping into the AI race? We've seen a few venture backed approaches where the idea is like instead of the Harvey approach, let's buy law firms. And we've seen even in the pre, about a decade ago, Justin Kahn was working on Atrium, this kind of like tech powered law firm where it was a law firm. But Harvey's made the choice not to, but what is your take on private equity dipping their toe into more venture scale opportunities and venture investors starting to look more at private equity style roll up deals?

1:46:11Yeah, well, so I used to, I came from private equity. So this is something I think about a lot. I think it makes a ton of sense and it's a continuation of the private equity play. A lot of the investments I did in PE only worked because you took 20 % of the cost out. And so now you have a much better tool to go and do that. But it very much is the playbook and it's what they're best in class at. And so do I expect that they'll be great at adding AI to the arsenal for how they get those margins up? Absolutely. I think that, you know, when I think of the businesses that I'm excited to invest in, it's like, okay, at the end of the day, it's who's creating gross profit.

1:46:45It's gross profit dollar creation. And you can choose to do that by investing in the billion dollar revenue company and taking their cost down 10%. Or you can choose to do that by backing, you know, that amazing DevTools founder that knows how to build like the AI native DevTools company that's going to create$100 million of gross profit off the bat. And so like, I very much personally like to invest in the founders that are kind of creating net new revenue dollars and net new gross profit dollars, but like multiple ways to play. I will say like, you know, I had to really retrain my brain when I went from private equity to venture and just, I mean, everything you, like the way you operate is just so different.

1:47:19And so I do think it takes a different type of culture to operate a roll up or a cost out strategy versus invest in startups. And so while I agree, like I agree with the strategy of PE firms doing their thing and venture firms doing their thing. I have a question mark on the blurring of the core competencies. How do you think about the different businesses that Sequoia is running right now from early stage, growth stage, beyond? How do all these things play together in the strategy? We've seen some venture firms even dip their toe into general catalyst buying a hospital network. Lots of people are thinking outside the box these days.

1:47:57There's the crossover funds. What do you think Sequoia it does best and what are you excited about in the future? We're not buying any hospital chains yet. I would say like if you think of our strategy, it's like it's seed to IPO and beyond for like the most ambitious entrepreneurs in the world. And so sometimes we're able to catch them early at the seed like Airbnb, like Stripe. And sometimes we catch them later on in their journey. But like the point of adding additional kind of pools of capital to our fund strategy has been when we find a winner in our portfolio, for example, take like a SpaceX, we want to be able to invest a lot of money behind that company as it goes on its journey.

1:48:35And the reason for the Sequoia Capital Fund is like even after these companies go public, we think a lot of that return is still to be had. And so we've modified our structure over the years to be able to support these companies as they grow and become later stage and go public. But ultimately, it's invest at the earliest point of conviction and and ideally that's at the seed makes sense do you expect to see more uh sam had a some interesting sort of uh quotes over the last week or so i don't know exactly when they were happening talking about uh the cost sam sam altman the the sam which other sorry sorry to the other yeah sorry sam lesson yeah yes um no he the the quote to To summarize, it was, or the line was something to the effect of the cost of AI or the cost of intelligence will just converge on the cost of energy or electricity.

1:49:32And I'm curious, you guys had Chase from Crusoe talk, and I'm curious if you think that that is a potential area that you expect to see more net new early stage startups exploring because it probably hasn't got enough. you should have nuclear and we're talking with this it was sean mcguire we were saying like there is no elon of energy yet but it feels like the last massive massive market that no tech founder has really gone and dominated in kind of the founder mode way we were talking about big oil there's still a bunch of huge companies can't name any of the ceos they're not really in founder mode they're kind of boring and maligned and it feels like there's an opportunity there but yeah sorry that's a lot no i mean it's a great question we uh we had chase from cruso on our on our podcast and i i will i'll put in the bet that chase might be that elon like figure he shared some stats that were amazing to me and it's like i'd always kind of thought about ai from the oh i can generate cool ghibli images uh perspective but i didn't realize the extent of like the sheer extent of the industrial build-outs that is happening to support all of that and so chase shared like if you look at like typical data centers today they're like 20 40 megawatt data centers like the biggest data centers of the world are in Northern Virginia or sorry, in the U S or in Northern Virginia, the aggregate capacity is four and a half gigawatts there.

1:50:49Chase at Crusoe himself has 20 gigawatts in pipeline right now, more than two gigawatts built out. And so like the sheer scale of the build out right now is just like nothing that we've supported in the past. And the bottlenecks are moving around. So like people, it's actually impossible to get chips now. A lot of that is easing and power is the new bottleneck. And so this is why there's so much happening in West Texas right now in Abilene, where they just have this like massive overbuild of renewables, especially with wind. And so like, I think very much you'll see a lot of the AI build outs following power and energy, because that ultimately is the binding constraint right now.

1:51:25It's a lot of sense. Well, we'll let you go. This has been a fantastic conversation. Come back on again soon. Went all over the place, but we'd love to have you back. This is so amazing. Come to our AI party next year. And I heard you were asking, Andrew, about swag. Oh, yeah. We have these amazing scented candles. I actually have one on my desk. Ooh, nice. We have these scented candles. I'll send one your way. Thank you. It was evidence that I didn't know our audience at all, but I enjoyed the scented candles very much. I love scented candles. Mother's Day is coming up, so, you know. I know. That's the perfect gift.

1:51:52Your segment's on the Himalayan Birkin. I was like, I love these guys. Yeah. They're going to be flying off the shelves. Yeah, yeah. After we do this show, everyone's going to go out and get one. Anyway, thank you so much for stopping by. Thanks for coming on. We'll talk to you soon. Have a great day. Bye. Bye. Next up, we got Will from Slow Ventures. I'll be right. The other side of Slow Ventures. Sam Lesson's business partner, obviously. He's been on the show many times. We had to swap him out. We're replacing Sam with Will from Slow Ventures. Very excited to have him on the show. I've been digging into a bunch of those questions.

1:52:26I still want to know more about the robotics timeline. I'm going to try and dig into that. I still need to know how. I need to talk to more researchers about how images in chat GPT works because I feel like there's something going on there. You know, you see it with the text models that there's very clearly, you know, certain filters running on top. You get these weird rejections with the images where sometimes it will, like just the Studio Ghibli thing is bizarre because Studio Ghibli is real intellectual property. Studio Ghibli is a real company. And when you say that, it doesn't say, hey, this violates our intellectual property rules.

1:53:05But if you ask it to generate a picture of Superman, it'll say, hey, that's copyrighted. And so I'm wondering if OpenAI did a deal with Studio Ghibli behind the scenes or something, or maybe there's some definition of how the IP shakes out. But hopefully going to have a lot more AI researchers and investors on the show to kind of answer some of the bigger questions that I have. But in the meantime, we're continuing to yap about venture capital with a venture capitalist. Welcome to the stream. Well, good to have you here. Boom. Are we live? Are we doing this? Oh, we're live. Yeah. Hit the soundboard every five seconds for this one.

1:53:41We need to raise the energy in the studio. We got to go. Venture capitalist yaps about venture capital. Every time we have someone We're slow. We go crazy with the Kyron. Expect some wild, wild Kyron going on. It's Friday, but we're not letting the energy go down. How's your Friday? I'm all good, man. I'm about to amp up. I got this. I got a few more things to get out. Then we got two Little League games this afternoon. So we're just getting going here. Fantastic, yeah. And we're coming up on the playoffs, so it's a big Little League game. Are you a screamer? Are you yelling at the coach? No, I'm not.

1:54:13The ref? I'm a coach. You are the ref. The other guys yell at you. I'm a, I'm a zend out coach. I'm a zend out coach. That's my philosophy. That's good. Anyway, what's up boys. Great to see you. Have you ever, have you ever been thrown out of a game? No, no. I had a dad who was a screamer. I had a dad. No, no. I had a dad who I loved dearly. We were best friends, but he was, he was a yeller at the, at the refs. And so I kind of swore that off a long time ago. Adapted, learned, improvised. When I coached, actually when I coached water polo, I did get thrown off. a game when I was coaching high school varsity water polo because I was being too sarcastic.

1:54:52I wasn't yelling, but I was too sarcastic and the referee didn't appreciate that. Anyway, you guys look great. I almost busted my suit out, but it wasn't for the next one. It didn't feel like my role. By the time we're done, all of Silicon Valley is going to be dressing in suits every single day. We're going to be switching to business casual. If you go to our website, you'll see in classic form, slow zigged when everyone was zagging. That's great. And decided to show our LPs when things are a little tight that we are extremely, we're not just serious capitalists, we're extremely serious capitalists.

1:55:25Buttoned up. A suit is one thing, but a tuxedo really shows people that you're serious about managing money well. Yeah, that's right. That's a great photo. Well, I've wanted to have you on for a while. Always enjoyed our conversations. I want to know what you disagree with Sam on. He's got a lot of hot takes. Probably everything. Probably everything. Welcome to SLO. What about his idea of like, you know, he's pretty anti-AI. He likes the AI cherry on top businesses. Are there any of these pure AI plays? We were just talking to Sonia at Sequoia. She's had a lot of luck finding enterprise AI application layer companies that are pure AI.

1:56:05They are kind of rappers, but they find these particular enterprise niches that can go really big. The honest answer is Sam, Yoni, Kevin, I have a really, we really enjoy pushing each other and using Twitter as internal slack and highlighting where we disagree. But the reality is, is we're all pretty well aligned. We say it differently. We get there differently. But no, I'm just as anti-AI as Sam. And I think, listen, the important thing, we're not anti-AI. You know what I mean? I think that's like the easy way to take it. Our answer is like, AI is rad. AI will definitely, what do we do with swear words on the show?

1:56:37Aloud? It's a family-friendly show. We won't cuss, but we also don't have a bleeping mechanism, so we can't bleep it out. I coach Little League. I can cut it both ways. Okay. Let's keep it clean. Swear like you're five years old. Dollar. Drat. Like, listen, I think we all agree. AI is going to break the economics of the world as they work in a massive way, right? So the problem with Silicon Valley, like, that's step one to a compelling venture capital thesis. And a lot of people just stop at that step. They're like, okay, cool. Economics are going to break. Let's start investing against it. And I think for us, once you click, okay, but where are those economics going to flow disproportionately, right, against someone that has a really compelling business model, right, that can get from here to there on very little equity?

1:57:23That's where our AI thesis breaks down. And it's hard to argue. You know what I mean? that a lot of the game, I mean, I think the hottest take that I do agree with of Sam's is this is not nearly as disruptive as people talk about. And it's more of an enabler. And the winners are going to be big companies with balance sheets and distribution and data and all those things for a long while. So I'm totally on board with that. That doesn't mean we're not doing things with AI, but I think they meet the next couple of click steps for us. And a lot of those do look like AI cherry on top businesses with great founders who get sectors and understand what's going on, right?

1:57:57And then they leverage AI in addition to a bunch of other stuff. At the early stage, there's been this pattern I've been tracking with the new generation of like Gen Z founders, essentially, where they need to break into Silicon Valley. It's really noisy. There's a lot of, I mean, honestly, the millennials are dominating, like the latest, you know,$100 billion company is Sam Altman. And, you know, a decade ago, it was Mark Zuckerberg when he was in his twenties. We have a new big power law winner and it's - That's an elder millennial. That makes me extremely, extremely excited and happy. Yeah, yeah, yeah, yeah, yeah.

1:58:33But the Gen Z entrepreneurs have been, they haven't had this massive power law win. Yes, there's scale AI, which is doing fantastically, but there aren't as many, like Gen Z hasn't really found their Zuck yet. And so they need to break through in a different way. And I've been seeing a number of young founders breakthrough through viral stunts on X and in the media. And when we talk to them, we often come away saying like, well, like maybe we didn't love the fact that they have to play this game, but it just feels like a game that they have to play. And I'm wondering if you have a take on like the requirement of modernity that you have to be such a showman now to get attention for your startup that you often have to push it really hard and make all these crazy claims and do all these stunts.

1:59:22There's a kid that we have on the show who's hiring 50 interns to do social media for him. And he's getting kicked out of Columbia and it's allowed him to raise money and it's allowed him to break into Silicon Valley. But there's always that risk that is taking you away from just going heads down, building the product, doing like the Dylan Field thing at Figma, where he was just grinding for years and then produced a great product. So are we in a new era or is this kind of just the natural evolution of breaking into Silicon Valley. I think a lot of that is like compensating for not actually having extremely interesting novel hypotheses that screw with the economics of the world and sick business model.

1:59:59So I, number one is like, I think it's copium, um, for the most part. I have another take that I've been wrestling with, which is like, I don't think we have founders that are capitalist enough. Um, I think like when you see Wiz print that outcome and then you find out it's like Sequoia, Doug Leone, a second time founder who was idea, like you start going like, oh my God, those are people that, you know what I mean? they play they play to win and they and they the scoreboard for them and everybody is in dollars and so i think that's like another thing that comes to mind when you talk about founders struggling to do that it's like i i've been kind of curious on like has there been an entire generation of founders on the back of hey it's product right that like aren't as capitalist as necessary and my my take on that is like there's actually three markets you got to back people or like the big rad companies end up coming from people that can do like are addicted to winning in three places the market for customers the market for talent and the market for capital and they have like a tremendous amount of interest and respect in all three games and they want to go maximize and manipulate all three of those i'm not saying the illegals like no don't go that but like they're looking at all three going constantly how do i move chess pieces to like win this game more and so um i think a lot of the stunts are are either from a product the oriented mindset of like, I just need people to see my thing and try my thing, which I just think asymptotes out at a certain point.

2:01:29So I'm not sure. I mean, listen, I think if you're really good at it, it is a way to bend some of those markets. You know what I mean? But I think you need to understand, I would say, and again, I don't want to talk about Sam last thing the other time. Our takes over that. No, no. He had a good point about this, which is like you sell AGI to raise free money for a consumer application. You know what I mean? And you sell self-driving in order to have enough equity to get the margin structure on your cars to a place that's sustainable. I'm like, there is a role for that, but it can't be the whole thing.

2:02:01Yeah. Am I interested in working strategy? I've been trying to figure out who patient zero was that like robbed founders of their like cutthroat capitalist nature. Um, and my working theory that I'm curious, I want to get more feedback on. I think Google was patient zero because they built a product and a business model that itself was like the most beautifully efficient capitalist thing that was like constantly operating at the efficient frontier all the time where they could come off of like, hey, we're just vibes and product. You know what I mean? And that was like the exception that proves the rule.

2:02:33You know what I mean? Because it was such an amazing – I mean I still – I live in awe of AdWords every day. Yeah. I think that's what I think about Google is 100 % right. They built a monopoly that just spit out cash and then they could just do whatever they wanted because they didn't need to be ruthless at all in anything else because they built the perfect money machine. And Morgan earlier was saying you could potentially say the same thing about Steve Jobs around the original, the product goat in some ways who didn't have to be as obsessed about the business model of the iPhone. So I totally disagree on that take, actually.

2:03:12And I think I – sorry, Jordan, not to cut you off. No, no, no. Go for it. We love disagreement. So no, so this – I didn't pick this up until recently, or maybe I knew it and forgot it. But I mean everybody knows like they founded Apple. He got thrown out when it was time to like run a serious business because he couldn't do that. Went away, wandered in the woods for reals, started next. And I didn't realize he plowed almost all of his Apple profits into Pixar. Did you guys read this article? No, no. I mean, I'm loosely familiar with that story. Okay, so not only did he do that, he ended up getting super deep in the weeds and running the game theory on their IPO and strategic investment from Disney.

2:03:55And he basically ran the most gnarly capitalist playbook there. He ran brinksmanship using the capital markets as a lever. And that's when he went back into Apple and crushed it. And so I actually think he like, is it this counterfactual? He's exactly the rule, which is you need to be on, you need to be on tilt in all three ways. Like what are customers, how do customers value? How do you win them? How do you win talent? And how do you win capital markets? And I, and I, when I read that story, I felt, uh, I felt like my cooked up theory is actually correct, which is like, then he re-enters, right.

2:04:29And Apple's a hundred thousand Xs. Yeah. I mean, I, I agree with, uh, that and I have like kind of a similar take, just the, the The difference between Apple and Google is that Apple does operate a little bit in the world of atoms. They have to actually make a thing. And so there is some sort of ruthlessness that comes from like the screws have to be screwed into the metal. And if you're not on time, you're not on budget, like things can really go wrong. Whereas if you just have this beautiful algorithm and this website that people just have as their default homepage. That gets better and more profitable with usage.

2:05:02Exactly. It is a little bit easier to run Google than it is to run Apple. Like Apple, like, you know, your supplier could kind of screw you. Like what is going to happen in Google, in Google's world? Anyway, I don't know. I mean, it's a good take. Anyway, that was, that's all right. You, you, you mentioned millennial founders and stunts and it got me on one of my nine talk tracks that I grabbed. Yeah. But I mean, we were just talking about this. Like there has been this trend and we were tracking it from, uh, yeah, they probably started at Google. Uh, but then, uh, Mark Zuckerberg drove like the accord for a long time.

2:05:30And then Sam Bankman freed kind of took that, Like, yes, I'm a billionaire, but I'm like the, I'm like the down to earth billionaire, the benevolent billionaire. And he was like driving a Toyota Corolla and it kind of, you know, created this meme around like, you can be really successful and wealthy, but like, you don't want to display it. And then the next generation kind of got caught up in this meme of like, uh, it's all about the experience. I just want experiences. I don't want any material things. And it's a big question about if you can't concretize what you want in life from a material perspective like you can't say hey i actually want a house for my kids to live in i actually want that cool car that i've been obsessed with since i was a kid can you concretize building an empire building a big company if you're not thinking concretely in materialist in materials like perspective maybe okay i just think the other people want to want to win you know what i mean they're just like i'm playing a game yeah i mean and he knows about zuck is like extremely competitive whatever he gets into it's like i'm gonna learn mandarin faster than anybody else that like so i i don't disagree that like i don't i i joke everyone talks about sam altman is this like visionary product he also drives a three million dollar car he like cares about you know what i mean he can't he knows the scoreboard's got the money um but it's a 20 million dollar car by the way oh he does there you go three million koenig's egg that's the daily i think and the f1 is in the okay well so on the young founders note something that i think is is happening is that so much venture capital appeared and became available to very young people that weren't necessarily exceptional in san francisco right let's blame this on will right people people that are people that are the smallest fund you've interviewed all week no so so so founders that are that are talented but not not the the sort of necessarily top of their class and if you give a young founder five million dollars and you tell them to run their first business they're going to act like somebody who won the lottery effectively they're going to spend money in a way that is just yeah give me a yc take i thought yeah and i and and i saw a founder there there was a no so there was a founder who recently raised around we've we've had him on the show i think he's super talented and he was like you know hiring 50 interns at once and then he put out a post he's he's going to hire a videographer this summer for like five half a million dollars a year and and i was like that's the one thing that yc gets right because they basically say spend no money until your thing is really working and then you can start pouring you know fuel on the fire but until your thing is working spend 20 grand a month right it's just been proven time it's really really hard to buy product market fit like you can buy growth post product market fit but it's very hard to just say.

2:08:18If you could buy product market fit, big companies would be doing it all day long for new products. I heard about another company that's planning to spend 20 % of basically two rounds that they've put together on a single launch day, effectively. And this is for a product that doesn't have any users right now. And so it's like stuff like that where it's like you're giving talented young people an obscene amount of cash and it's the same thing that would happen to you know somebody that's scratching something off in you know uh uh you know at a gas station and they get five million dollars and they blow it in a year kind of but it's almost worse than that which is like hey here's a bunch of money that was like pretty easy to come by and we don't actually want to hold you accountable to the money coming back because it's product and it's almost we fell on this trap of and you know i'll get on my rant about value proposition but we fell in this trap of like backing people to do like subjectively valuable things and that's like a really hard capitalist endeavor to scale which is like we we so we talk a lot about value proposition existing on a two by two of quantitative and qualitative causal and correlating and you can build great businesses in all quadrants but like what you do to scale them out of the gate is totally different and i think we got down this like hey what's the value of that product it's like i don't know what's like it's different to me than it is it's like Like, handbags are clearly valuable, but, like, why are they worth$10 ,000 for some people?

2:09:46Like, you can't, there's no math, science, logic, or anything behind that. It just is. And so I think, Jordi, it's actually another thing where, like, here's a bunch of money and go do things that, like, do not have. We talk a lot about our job and our money is to figure out, does something work, right? Like, with a true or false answer. And the problem is a lot of things got started that, like, don't have that true or false answer, right? So it becomes about derivative signals that actually aren't, like, connected to building killer businesses. So a yes and to your comment. Yeah. Sam's been really big on AI as a sustaining innovation in the Mag 7, in the big tech world.

2:10:23At the same time, it feels like the big tech, like the stocks are performing very well. The financials are fantastic, but the products seem to be faltering. You can't find the right Gemini app for Google very easily. Apple intelligence, no one's really raving about it, at least in the tech community. What is your take on big tech? Is it a better time than ever to start a company that takes a shot at a product that normally would be, you know, owned by big tech? I still think that if I were running Google, well, I don't think 90 % of the use cases in AI are actually like interestingly monetizable.

2:11:01I don't know that it like – and I'm starting to watch my own usage now and question that a little bit. But like the majority of the content being generated from AI and the majority of my usage is like not average. I don't know that I'm like that stressed. You know what I mean? Like what am I – I'm like going into Chattapiki and talking to them about what I'm going to talk to you guys. I didn't do that. But like that use case is like really great, really valuable to me. what's actually the back then back to subjective value prop like what's the value of that very hard to put a number on and very hard to monetize so i i i was on i tweeted it a while ago but like someone did i forget who did someone did a really good like take on this which is ai is going to be awesome to change the world and adwords is mostly safe from that so i and i don't i just i don't like i don't love the hey i want to take on the product because they're not good you know what i I mean, I think problem statements when it comes to building companies are not nearly as powerful.

2:12:00They don't do important companies nearly as much as like, I just believe the world works in this way. Nobody else thinks it works. And if I don't work on that, somebody else. And I've been big on like, will the future? It's like, no, no. Go work on things you think are like just truths. You know what I mean? Because you don't be pissed if somebody else works. So I don't have a good answer to you on like you take on Google right now. because I think the answer has to be you have to like come out and think the world works dramatically different in 15 years and start building against that can you give us an update on on your guys's franchising thesis uh these sort of um for I feel like uh the venture world got excited about four wall businesses because there's there had been a bunch of M &A over the last few years and people realizing like hey you can kind of spin up a brand and you know prove it out a little bit and get a bunch of other people to, you know, scale it.

2:12:54And I feel like this idea of, you know, this business in a box has been very prevalent for the last few years. You guys have had, you know, your own thesis around it. But I'd be curious to get an update there on where you see that kind of category and how you see that opportunity today. Yeah, I mean, top of funnel has been lighter than we wanted. I think the overlap with people who think this way and like, think of slow is not so that's that's something we're constantly working on is like seeing more in that space um listen the way we got to that space is like what are the raddest businesses out there you know what i mean just from a year in year out efficiency efficiency on equity just like what are the things you would want to own for 30 years and hand off to your family which i think is like a really great framework on this stuff yeah and you kind of quickly get to franchises you know what i mean like they kick off a ton of free cash flow pretty durable for the most part really efficient scale dot dot dot so like that was just always my interesting take is somehow venture capitalists became software investors and not uh and not folks looking for novel hypotheses that have killer business models attached to them um and so anyway that's how we ended up there how's it gone it's like really fun i think it's a very logical output for a lot of innovation um i don't think we're doing nearly as much as we would like there but we continue to look every single day and again for us for us it exists on a like the it's it's like one expression of you created some novel ip that has rad economics you know what i mean that like does something that creates a ton of value in the world um and so we talk a lot internally we're spending a lot of time on our our gbo thesis our growth by buyout thesis that we rolled out probably you know maybe first i think we did but uh with some friends in our in a company called metropolis and it's like i'm not wedded to that.

2:14:45I just thought that that you know what I mean? I looked at vertical SaaS and I was like, Oh my god, it actually came to me if it's interesting. And industrial rail, I got pitched by these guys with a sweet industrial rail logistics company. And I was like, I love this, I want to be involved. Right. And then I started digging in and realizing there were some tough go to market dynamics where you might actually double the earning power of an entire industry, but be an 8 million ARR company. Wow. I mean, directionally correct. I'm making that up. But I was like, That's wild. You know what I mean?

2:15:12I was like, how did we get to this place that you could actually do something wildly disruptive and everyone else but you gets paid for it? And that's where I started peeling away the layers of like, right? So like, it's like, oh, what if you bought a railroad and you double its profitability? You'd make more money than an 8 million air or a SaaS company. right then you go what off it yeah right then you go to one of the other versions that it's like oh maybe some some places it's more efficient to vertically integrate maybe some places business in a box and then a woman who works for me is much smarter than i am when i was spouting all this off quickly was like what about toast and i was like yeah toast should definitely be a software company so that kind of informed our framework that franchises slot into how do you think about buyouts um i mean there's venture capitalists that are getting in and on the equity side there's some funds, private equity funds that do equity investments and they have bank partners that do the debt piece.

2:16:03I imagine that you don't have a credit fund separately that's managing that, but is that coming? Okay. And then there's also like the private credit guys that come in with just the credit piece. How do you think that that's going to evolve and is it important for VCs to kind of have at least partners in the whole capital stack? You'll have the partner. I mean, like so again metropolis is our kind of lighthouse case study in this if you look at i think the where there's a hole is kind of the growth equity portion of this right so we'll find someone to go do some breakthrough product work and go oh my god the economics are changed and we'll fund them to do a small scale buy of a business and even thought and see if you can like there's basically two experiments is the product actually transform economics and does it translate when you own and operate the p l then there is a hole on the equity side for like great let's go buy 10 of even not not one.

2:16:54But if you look at the Metropolis deal, it's Wall Street. You know what I mean? So I think I had enough evidence, this all slots right into their models and Vista's credit fund comes in and da-da-da-da-da. I'm curious. We had Harley on from Shopify yesterday and it feels, you know, in many ways, it's the most significant SMB platform in the world. It's a product that at scale entrepreneurs uh it's so critical to the business that yes they're going to have issues with it but generally people feel great about shopify and i've been interested to think about in a lot of other categories you have these sort of like jiro dreams of sushi type opportunities where linear was able to uh look at everybody hates jira but like we need tools like that And so what if you just built an amazing issue tracker from the ground up and really cared about the craft?

2:17:50And if you look at Shopify, you could look at, you know, oh, this design thing's not great. Or what if this was better? Or what if we, you know, made this more intuitive? But I don't believe that you could take the most talented people in the world right now and try to get them to rebuild, you know, a commerce platform like Shopify. and just given the developer ecosystem and now shop pay and all these other things i don't know i don't think i'm struggling to see how somebody replaces shopify in the market right now and i'm curious uh how you know if you guys have looked at any other any other players that is anybody even daring enough to try to to take on uh shop i i just i'm with you i think it's It's a fool's errand.

2:18:40So back to like value prop, which my whole North star is like, I need to buy someone that's creating a ton of value and can capture a lot of it pretty efficiently. That's the end of the day, what we're trying to find in the world. But value prop has to sit on two axes. And again, so this is like, everyone calls me Professor Will and you're hearing why I go in some long tangent that's like all theoretical. Love it. But you gotta have an absolute value prop, but also a relative one, right? Like if you go out, great, you create a dollar for people they're stoked. if the other solution creates like 98 cents of value right and it solves the same you guys are going to be in a dogfight from a sales and marketing standpoint right you're going to you're like you yes you create value but you're communicating that to the market is going to be super hard and so i think the issue with taking on someone like spotify or shopify is the value like it's hard to create relatively more value can you do things nicer on the edges yeah but like the core functionality of like, I have a business that generates profit for my family.

2:19:36Yeah. It's like, that's a lot of it. And it's very hard. The switching costs are super high. There's a ton of risk. And it's hard to do that dramatically better and be like, Hey, just by using our platform, you know what I mean? You will get, make dramatically more money. And so. And Shopify can say, to be clear, they can say, look, if you have shop pay, you're going to get an incremental, you're going to get an incremental 10 of revenue just by default because the check you're going to have less abandoned carts we lose you well he's frozen let's kick it over he's absolutely ads go to get bezel.com your bezel concierge is available now to source you any watch on the planet seriously any watch let's bring in some more soundboard uh we can also sing the wander song find your happy place find your happy place book a wander with inspiring views hotel graded amenities dreamy beds top tier cleaning and 24 7 concierge service it's a vacation home but better folks can you imagine if if slow is just so committed to just being lean like their companies that they didn't pay for Zoom.

2:20:48The internet bill or something? No ISP? Anyway, we have three founders coming on back to back to back. Tell Will he's happy to jump back on. We can close out, but we don't have anyone in the studio right now. But we are going to talk about Nourish, Matic, and Fastino. Three wildly different companies. One in the healthcare space. One in the... We actually have the robot in the box over there. We should unbox it on the show with him. It's kind of a next generation Roomba talking about cleaning in your house and very excited to talk to him about that. And then Dusty is training AI models on commodity graphics cards.

2:21:30So I'm sure there'll be a bunch of interesting things that we can dive into on how DeepSeek was able to train on low-grade, unoptimized cards, if they can do it on gaming quality cards. I'm sure other people can do it on all sorts of different cards. uh should we do some timeline in the meantime let's do it let's talk a message will i i'm i'm actually uh concerned hopefully like you know oh we have someone okay great yeah let's bring him in we have aiden welcome to the stream how you doing welcome what's what's up guys thanks for having me uh congrats on all the success you're having with the show you guys are blowing up yeah i mean success uh you know technical difficulties it's uh one thing after the other today on the stream but we're doing well.

2:22:12Our last, our last bounce bounced in two seconds, but it was a great conversation and excited to have you here. Would you mind kicking us off with a little bit of an introduction on yourself, the company and the news from this week? Sure. Yeah. So I'm Aiden Dewar. I'm the co-founder and CEO of a company called Nourish. Nourish connects chronic condition patients of all different types with a registered dietitian over telehealth. And we get it covered through your health insurance. We have the largest network of dietitians in the country. We have over 3 ,000 dietitians on the platform. We've served hundreds of thousands of patients, and the vast majority of patients pay nothing at all through their health insurance.

2:22:51We're in network with most major commercial insurance payers, Medicare, some Medicaid plans at this point, and have made a lot of exciting progress that we can speak about today. And then, yeah, in terms of the news you mentioned, we announced a Series B round recently. it's a$70 million round. 70? Let's go. Fantastic. Yeah, ring the bell. Ring the bell. Congratulations. Thanks. Yeah. So, you know, it was led by JP Morgan's Growth Equity Fund. And then participants from all of our existing investors. So Index Ventures, who had led the Series A and Thrive, who had led the seed round. And then Box Group, who I know our mutual friend, David Tish, who's invested at the seed round, has doubled down at every round, including this one.

2:23:37We got mutual friends all over the place. James Pierpont Morgan, doing him proud. JP Morgan in the deal. Is this, like, why was this not covered by insurance before? Is this a regulatory change? Is this technology? What's driving the growth in the business? Are you just the first to think of it? Or is there something unique that's allowed you to start covering this type of treatment with insurance? Because that seems like a major, major unlock, right? Yeah, yeah, good question. So this actually first started getting covered in the early 2000s by Medicare. I think it was 2005, Medicare had done a study and seen ROI for this sort of care.

2:24:17And really the insight that Medicare put together, which is kind of the same one we put together, is that you look at the healthcare crisis and trillions of dollars to spend and hundreds of millions of people with chronic conditions and millions of deaths each year. actually kind of you start double clicking, most of those are downstream of nutrition, and the vast majority of spend is just downstream of a handful of conditions, you know, obesity and heart disease and blood pressure and cholesterol and kidney disease and, you know, a few others. And so they had seen ROI for working with a dietician, you know, it's a really effective intervention and started covering it.

2:24:47And then the Affordable Care Act 2009 carved it out as a preventative benefit. And so that's this coverage, and then kind of the commercial payers, you know, followed and more Medicaid plans. And then kind of the more recent unlock, which is right around when we were starting the business, the business a little over three years old was telehealth coverage at parity with in-person because of COVID. And so that was kind of, you know, maybe the biggest, you know, recent regulatory unlock that, you know, made this, you know, a lot more accessible. Of course, you could have built this business, you know, in-person, but would have been much harder than the way it's built today via telehealth.

2:25:17Yeah. On the subject of the raise, how is it different raising from, you know, major financial institution versus a, you know, typical Valley VC, I imagine it's both similar and different at the same time. Yeah. So I think actually, you know, in terms of the actual firm, the person we're working there, Paris Heyman, who we're excited to partner with, he actually was previously at Index Ventures and been part of the team that run our Series A. So actually, you know, wasn't so different in practice in terms of just, you know, we'd already known him. And I think the, you know, the processes are similar.

2:25:52I think the biggest difference really was obviously at each incremental stage, there's different expectations. So series B, very different in terms of maybe more focused on the actual metrics and getting really deep on the data than series A or seed. And so that's definitely an adjustment, but we're fortunate that the basics of the process and the buy-in on the mission and vision and stuff like that was pretty similar to last round given we already knew Paris. That makes sense. What is the tech stack like for one of these companies? Is there like a plaid of insurance that you can kind of build on top of like or are you writing integrations and is this uh is there a piece of like you know service as a software where there's a lot of humans in the loop in the short term and then in the long term it'll be ai driven like uh what does the actual uh build up of the company look like uh so there is not not a not a ton of what you said in terms of the insurance contracting piece that's something we've and kind of a big on luck and the value we provide for dietitians is actually going directly to these insurance companies and negotiating these contracts.

2:26:53And I think that's one of the reasons most dietitians didn't accept insurance historically is it's a pain to go directly to one of these really large enterprises and negotiate a contract. And then even if you get the contract, deal with the ongoing administrative burden, licensing, credentialing, billing, and whatnot. And so for the most part, we've built all that, whether it's from a tech perspective or just from an you know, operational competency perspective in house, we have leveraged, you know, a great vendor called Candid Health for billing. And it's kind of, you know, API forward billing platform that we've built on top of.

2:27:22And so that's been a big unlock on the billing side. But I would say most of the other stuff from a, you know, contracting and credentialing and licensing perspective has been, you know, a lot of manual work and kind of competency that we've built up in house. Yeah, can you talk about trends in the evolution of dietitians and how they work? We There was some research that showed that artificial intelligence was particularly good at therapy. And you can imagine the future is, you know, you just have an LLM giving you diet advice. John, of course, they're on TikTok. They're like, OK, all my clients are keto now.

2:27:55And the next week they're on TikTok again. They're like, OK, we're going paleo, everybody. We're going carnivore now. How's it involving? No, yeah. I mean, you're hitting on an important question, which is like, you know, we've seen AI be a big unlock for the business. And I think, you know, you're kind of asking more about, you know, the patient side, which what we've started to build in into the app is, you know, essentially a lot of tools for the for the provider to use or the patient to use without the prior in the loop to get a lot of the, you know, info you're talking about. And the way I think about it is like, you know, about 80 % of the equation is actually behavior change, but about 20 % and like accountability.

2:28:32And that's where the human is, is especially good. But 20 % is a little bit more like educational blocking and tackling. And Google is not especially good at that. but the LLMs are. And so that's a lot of where we're leaning in is like, how do you give the dietician tools with AI to really provide amazing care on the things that only they can do and then leverage AI for the rest of it. If you've gotten to the point where maybe you're obese and have multiple chronic conditions, it's not that you're just kind of a couple of educational tips away from having success. You've typically built up these habits over years, if not decades.

2:29:06And so while the LLM can provide some help in terms of educating you on what's good for your diet and personalizing it and whatnot, the dietitian is really important for having the human in the loop and really the accountability to do that behavior change. And so as the LLMs keep getting better, we want to really lean in there. We think we're really well positioned to leverage AI in this way given we've seen hundreds of thousands of patients and have a lot of data on what works in terms of driving outcomes and habit change and behavior change. And so we really think kind of the synergy is having both the dietitian and the uh and the ai and the loop kind of drives better care ultimately very cool how what have you seen around the adoption of glp1s uh the concern that i've always had with them is that people that should just focus on their diet end up you know taking you know a magic shot that makes them lose weight uh and that's good because for for a lot of reasons but um uh i have to imagine a lot of the a lot of the end patients that you guys work with uh are both using glp1s and you know thinking about um the diet uh uh very intensely but what's your reaction been and i'm sure that's a question you you've got a lot even during the the fundraise i imagine yeah yeah for sure and um it it has been really interesting to see the you know the evolution of that over the past few years and you know the our dietitians had a lot of expertise working with glp1s even before you know the the recent explosion, you know, I think, as you all know, they were originally approved for diabetes.

2:30:39And so dietitians have been, you know, working with diabetic patients on them for a long period of time. But of course, the, you know, the recent explosion that everyone has seen for weight loss has also affected us. And what we've seen is that they are a really, really important innovation are really valuable for a lot of patients and are really useful, you know, tool in the toolkit. And we've definitely seen it skyrocket in terms of our patient population and people using that, but they aren't a panacea. So, you know, there's a lot of side effects. A lot of people fall off the medication because of the side effects or either just like, you know, the logistics of adherence of getting access to the medication and injecting yourself every week.

2:31:15And so what we've really done is kind of pair ourselves as a complement to the drug. You know, these drugs are FDA approved to be paired with diet and lifestyle change. And so we see ourselves as that part of the equation. And, you know, there are a few different ways. We've kind of built out our care model across kind of the journey of a GLP-1, which is first, you know, we have a care program that's before you even get on a medication. So a lot of folks want to, before they try a medication, see if lifestyle change will work for them. And so we'll kind of have a, you know, a step therapy type product where you try it before you get on the medication.

2:31:44And then if you do choose to get on the medication, we have, of course, a commandion program that will maximize the efficacy of the drug by mitigating side effects. And, you know, we've seen a lot of cool outcomes with people losing, you know, about 33 % more weight with the dietitian plus the GLP-1 than they do just with the GLP-1 alone. And then the final piece of the equation, which is, you know, maybe the most important, which is actually getting off the drug. You know, a big problem with the medications, which are, you know, our payers see, but also our patients, is that when you get off, you often see rebound weight gain.

2:32:12And then, you know, of course, for the patient, that's really frustrating and difficult. but also for the system and for the insurance payers, uh, from an ROI perspective, now you've spent all these dollars to have someone lose weight and then regain it, which of course is, you know, uh, worse. And so, uh, we have, you know, a program built around getting off the medication and making sure that's, that's sustainable for the long run. And so, yeah, we've seen it be, you know, a really big tailwind for the business because, uh, it's top of mind for payers in terms of managing this cost and lifestyle change is a part of it.

2:32:37And then of course, patients are really motivated to, uh, uh, change their behavior once, once they're on one of these medications. What's the use of the funding? $70 million, a lot of cash. Is that R &D spend? Are there growth channels that you're investing in? How are you thinking about growing the team and the company over the next couple of years with that new Series B? Yeah. So, I mean, the way we've always kind of thought about our success is scale and outcomes. So it's, you know, how many people can we help and how much do we help them? You know, the quality of the care. And so I want to use the capital on both.

2:33:09So, you know, scale, maybe most obviously, we want to really invest in expanding our network of dietitians. So as mentioned, have 3000 today, you know, the largest in the country, but I've had a long wait list for a long period of time of dietitians who want to join who haven't been able to. And so really want to expand our network of dietitians quickly to be able to support more patients. Of course, on the patient side of the equation, we'll invest and go to market to get in front of more patients to raise awareness that this is something that's, you know, effective and covered by their insurance.

2:33:34And then on the payer piece of the equation, want to keep investing in partnerships with insurance companies. As I mentioned, we have pretty good national coverage, a couple hundred million lives covered at this point, but there's a lot more plans we can get a network with to expand coverage and want to continue doing so. And then maybe last, but certainly not least, maybe the part that I'm most excited about is really accelerating development and product development effectively. I think we talked about some of the AI use cases we've seen for patients and dieticians. And I think basically everything we've built there has been successful for those parties and driving outcomes and saving dietitians time and, and, uh, patients really love it.

2:34:09And so, you know, we're really eager to continue investing in a lot, a lot of that, you know, product development and, um, you know, long-term, I think the reason we got into this, I had had my own chronic condition. I had really bad migraines and I'd worked with a dietitian to solve it. And, uh, you know, of course dietitians, uh, were really valuable for me. And that was part of the reason and my co-founders, and that's part of the reason why I wanted to expect, expand access to it. But we know that, you know, part of healthy lifestyle change is not just nutrition, but things like mental health and sleep and fitness.

2:34:36And I think there's a lot of interesting kind of applications we can do of that in our product to drive better outcomes. And so really want to run at that pretty quickly. What else are you seeing in the overall health tech market that's interesting or complementary? We talked a little bit about GLP-1s. There's a lot of online pharma companies that are working in that market. But where else are you seeing exciting companies or trends that could be complementary to what you do? yeah i i think they're kind of on the on the vectors that i i just mentioned so of course there's been a lot of movement in the in the glp1 space and we've partnered with uh with players there to to help get access to these medications to our to our patients um you know a lot of our patients work with a you know a therapist in tandem and so you know there's you know great companies in the mental health world that we've we've referred to um i think labs are are really interesting you know about 85 of our patients get you know labs done each year it's of course a natural part of the care journey of both getting a baseline of you know discovering you know where you're at in your health journey and where you need to improve but also as you're making progress over time um it feels like the lab was just destroyed by the theranos story and no venture capitalists will touch it now um but uh yeah i mean it does seem like there's a lot of new new entrants there superpower and function health kind of doing uh the the upmarket version but there's lots of other ways to get labs done, obviously.

2:35:57Yeah, and it's, I think, you know, of course, Theranos was, you know, more about the actual kind of base level of like getting the lab work done. I think a lot of these, you know, players that are doing cool work that you mentioned are more about kind of facilitating getting the labs through the large lab companies. And then what AI is really great of like interpreting the labs. And so, of course, a lot of our, you know, our patients are getting labs done consistently and think the people there are doing a lot of cool work. That's cool. Jordy, anything? Amazing. This is great. This is fantastic.

2:36:22Thanks so much. And congratulations on the massive round. Say that again. Thanks, guys. One last size, guys. It was great to be on. Y 'all ask good questions. And I was thinking about it as I was getting on. Feedback's really big on our team. And I always get feedback about talking too fast. And so as I was going on here, I was like, oh, I got to make sure to talk slowly. But then I realized YouTube and podcast ad, that's probably where most of your views are. People just change the speed they want anyway. A lot of people say they just listen to the show on 2X because it's too much content. The real high performance on 3X.

2:36:54That's the future. People can listen to whatever they want. Anyway, thanks so much for stopping by. We'll talk to you soon. Great to meet you. Cheers. Next up, we have Matic Robotics coming into the studio. I'm pumped for this one. Yeah, yeah. You've been talking to the founder. I was DMing with the founder. He sent us a robot. We're going to have it be cleaning our new studio in no time. We're very excited for that. Wired said, this is the best robot vacuum we've tested. Fantastic. It scored a rare 10 out of 10. Let's hear it. Bumble, bum. Can I get the Ashton Hall sound effect? There we go. I haven't heard this at all today.

2:37:26It's killing me. Let's go. Bring him in and play that effect again, Jordy. Welcome to the stream. Welcome to the stream. Congratulations on the Wired article. Congratulations on all the progress. Can you kick us off with a little introduction? He's got robots and homes. Robots and homes, yes. They said it could have been done. They said it was a 2035 thing. Most people, they ship a render. They ship a video. This guy ships a real robot. What a concept. Welcome to the stream. how you doing thank you guys for having me appreciate it um i'm a president co-founder at matic we build home robots and my background is in computer vision and product and we prior to this we were at nest so that's a quick background and oh that makes sense yeah okay so yeah how many people thought you were crazy to go make a you know another take on a on a you know the next roomba you know whatever whatever the pitch was i think they still do uh i think it's still why are doing floor cleaning robots that's the that's the question we get quite a bit and and the answer really is that this is the only robot with scale yeah uh this is the the irony or our of what we are doing is that a thousand robots ship we are already the second largest american consumer robots company wow so we talk about this point of view that there is this perception that navigation and mapping and localization in indoor environment is a solved problem.

2:38:49But I tend to think of it as a Fermi's paradox that if it is a solved problem, where are all the robots in our lives? Why aren't they at the airports? Why aren't they at the airports, grocery stores? Why aren't we swung with it? And the answer is that it's actually quite a hard problem, number one. Number two, economic viability, which is making it profitable and surviving as a business itself is a challenge. And then on a flip side, making it valuable for customers. Yeah. Talk about the evolution of the relevant breakthroughs in artificial intelligence and which ones you are a true beneficiary of.

2:39:26ImageNet, obviously very groundbreaking. The transformer architecture. Haven't heard about that having an impact, at least in the Amazon's Roomba context. Are you using transformers? What about LLMs? What about diffusion models? Matic uses a vision-first, vision-only approach, which John has been nerding out about for months now. So I want to know, does this lead into the why now? Or how do you tell the story of the underlying technologies that have led to an improved experience here? A great question. We actually left Nest and Google in 2017 to start working on it because of two trends. One is a self-supervised learning techniques that were emerging, which is what LLMs are essentially.

2:40:11They're learning on their own. And then second one was that my co-founder and CEO Namneet helped spec out Google Coral TPU from Nest's perspective. So that trend of AI chips coming out and compute skyrocketing was a trend we saw coming. And between those two things, we thought it was possible to build edge device robotics. And the reason we thought edge device was critical was we as humans, we don't have hard minds. latency is really critical, especially in a dynamic environment that we live in. So we always thought that robots have an ability to make decisions really fast. And specifically for indoor robotics, we just felt like indoor world was built by humans for humans, for our visual perception system.

2:40:50So vision only robotics was the only way to go, that it needed the same perception system as us. If you're trying to build, let's say, level five robots, well, level five robots for cars means that cars drive like humans. So inside home, it means that they behave like humans clean like humans manipulate like humans so it should have the same perception system and those were the trends that helped quite a bit and and in doing that and now um if you look at a robot what it does is it builds a google street view like a map on its own so the way we thought about it is we as humans we go into new home new environment new indoor space we self-explore self-map and then remember exactly where we are so localized can robot do that well the answer is Yes, it can.

2:41:31Our robot does that, but it still has the same ability as cats and dogs. We can't tell our cats and dogs to go sit by the couch or go in a living room. They don't understand that yet. And that's where VLMs and some of these open source, Dino and Clip, some of the models that are being released are really useful because now we can extract semantic embeddings and information at the image level out of it. And we can actually transfer that into our map at a voxel level. So each of our maps are built using voxels, which is like a 3D pixel, one centimeter by one centimeter. So each voxel actually knows that it belongs to a chair or a human leg or a child or a piece of furniture.

2:42:11And that's when you can start asking and doing all kinds of things like, hey, go clean by the bookcase in a living room. And it knows what you're talking about. So it's really that's the next layer, which is turning into much more of a natural language interaction between a robot and a human being. Talk about the actual training runs that go into your models. Is there a concept of iteration on the training runs like GPT-1, GPT-2, 3, 4, 4.5? He said they spent less than a million dollars on NVIDIA chips. Okay, yeah. So break that down. How do you think about, obviously, you're acquiring data constantly, but then is there this pace of let's do another bigger run?

2:42:52And is there capital at risk when you actually roll out a big training run? Great question. So I'll take a little bit of a higher view and come down. But there is this concept in self-driving car as well as robotics that there would be one God model and that God model would do everything. When you look at practicality of a deployment, almost always there are multiple models at a smaller level that you do it. So we've always taken this approach that we're not trying to do research. We're trying to build product. So whatever is available to us, let's go do it. So our approach has been a combination of, obviously, neural nets and some of the work that's happening.

2:43:29But also what we do is referred to as a special AI. So that's a term that Dr. Fifi Lee really prioritized. And we use human information bottleneck principle. So the way we do this, we have image-to-voxel neural network. And then we combine that with long-term SLAM using both classical and techniques and build this world. and then the physics of the world is permanent. Now, as a human being, I can know that there is a wall here. I know what will gravity do. So based on that, for us, it doesn't take 26 ,000 iterations to learn how to tie shoe places. So in the same way, once you know the physics, you can predict things of that physics that I know certain objects will topple over if I were to do that.

2:44:11So that's how we think about it. So we use information model-like principles. So for us, computes and data has been critical, but it's less of a traditional logistic gigantic data set and let a robot do everything on top of it. So it's less of a compute intensity, but there is obviously iterations. Is there something like a mixture of experts model that you could kind of pull from and design to kind of scale up the model? Is that relevant at all? Absolutely. So we have our own master student models and stuff. Exactly. And the way we use it is that, hey, the 3D part of it, we use traditional techniques of a special AI, then adding semantics and understanding and context part of it.

2:44:58That's our master and student model works very well. But even for the neural net, sometimes for the precision, there are master and student models that we can use for precision as well, just to see the way humans see. So occupancy network that we have, a lot of it is inspired by the approach that Tesla has taken over the years to build their full self-driving. Talk about simulation. Are you using a lot of simulated data? I imagine that you could procedurally generate a million or trillion households with different furniture, legs and stuff pretty easily. Drive virtual robots around that, use that to generate data.

2:45:30We've heard about a lot of that in the humanoid context. Are you seeing luck with synthetic data for your product? Absolutely. I think we started with self-supervised learning. Then we realized that simulation and supervised simulation actually works very well as well along the way. So we built our own simulation environment using Unreal Engine with our own robot. So we've customized it over the years and have a large set of environment. But that usually takes it, what we've seen is that it takes us to about 0 to 80%, but that the final 20 % always comes with the real-world data. So we initially trained to create, quote-unquote, the master model, But then the precision and fine tuning almost always comes from adding real world data.

2:46:09Switching gears a little bit on product strategy. I imagine you have ambitions well beyond this initial form factor. Walk us through. Yeah, walk us through maybe the history. Was this always the form factor that you were going to start with? And at what point do you look at kind of expanding from here? Great question. So we always imagined the goal was always to go build Rosie the robot. All of us want that. Something sort of Alfred that comes in a home and just takes care of everything. But we thought that the best way to do it is the way human child grows, which is in the first five years of human child, they just learn how to navigate from a perception perspective.

2:46:51They're just trying to make sense of 3D rules, and they pick up the object and learn that it drops. So in the same way, floor cleaning robot allows us to do that. But then we evolve and say, just like a 5 to 10-year-old child, Can it start picking up shoes and moving it around? Can it just organize unbreakable items? So for five to 10 year old, we don't give them knives and scissors and all the risky stuff. So in the same way, can it start with this task and can we along the way productize it and start shipping and then ultimately put it all together as a full blown robot? And we thought this approach was better because as we did consumer research and we always start with customers and work backwards, we realized that there is a lot of apprehension about robot and whether they can do things accurately.

2:47:30So even though, you know, robot vacuums have been around for 23 years now, they've only penetrated 13 % of the U.S. households. 87 % don't even have it yet. And the reason is because they're just not that good. They're not accurate. They're actually kind of dumb. So for 2002, they were an amazing device, but they hadn't moved forward. And we thought that purpose-built device that solves the problem to the nth level is a way to earn customers' trust and then iterate to the second, third, fourth task. And the way we built it is if you see our current robot, we have a black crown. We have a black border on the top.

2:48:06That's where the eyes and the brain sits. And it's very much like a human being. And we always thought that we just have to build that once and then it just grows up just like a robot. So as a robot scales, it will scale. So the beautiful part of what we've built is that and then we have demos of it in our thing where we can just raise the robot, maybe even put it at a six feet level and everything just works out of the box. So you can put it on top of the humanoid and it will map the entire space for humanoid. And it was six degrees of freedom with the same precision at one centimeter level. So are you thinking about adding like a robotic arm so it can pick up a shoe and put it back in the closet?

2:48:39Absolutely. As we're going to users home, I've been in about 200 homes now, real homes now. And parents always talk about, can you just give me a tall cleaning robot? That's my biggest pain point. So that comes again and again. And the second thing we've heard is a lot of kids talking about their parents who live on their own in Florida or Texas, and they don't have a time to go there. And they're not technologically savvy as well. So they're like, we'll take this robot. We'll actually control it, clean their home. But can you also send me a 10-second time lapse at the end of the day to confirm that they're okay?

2:49:13How do you think about privacy? I imagine you have to message like, hey, we need the data, but we're going to anonymize it. You can trust us. but then there's data leaks that happen. I'm sure this is an important part of your messaging, but what are you saying to people? Huge, huge part of it. So prior to, we worked at Nest. I was a product lead for Nest Camera, so I know privacy is a big deal. So that's why we do the whole thing on their edge device. And then the way we do is just opt-in. And we always knew, and this is our prior startup at Flutter, we also did gesture detection where webcam and did this, that if you build a trust with users, there will be a spectrum of users.

2:49:46On one end, there will be users who says, don't ever take even my telemetry data. I don't want to share anything. On the other hand, there will be users who says, take everything you want. I don't really care. In between, there are lots and lots of users who would say, hey, these are the long tail at which your robot fails, and I actually want you to help get better, so we'll share the data. So we haven't even done it automatically, but there is a record button on our app, and users have already uploaded thousands of hours of data with permission on their own. Do you have evals? like like when you train a new model do you put the robot the final eval is can the matic robot clean up after two toddlers after they've had dinner i was gonna say frat party you know throwing frat parties toddlers toddlers are more messy they make a bigger mess spaghetti and meatballs a regular vacuum can't even no no no instant instant disaster it needs to sleep and all sorts of stuff anyway jordan last question how do you expect the humanoid how do you expect the humanoid market to play out over the next few years, right?

2:50:48It's obviously an area that you guys want to, uh, you guys will be competing in that market over time, but clearly made, um, some, some big decisions around how to get there. Uh, there's a lot of companies that have raised so much money that they, and, and, you know, once you've raised a couple billion dollars, people are going to want you to be shipping or at least having, you know, robots that are, that are creating value uh in in these settings um but at the same time like iRobot you know and we talked about this offline iRobot is the biggest robotics company in the u.s with 50 million units shipped amazon robotics is the second at 750k and then boston dynamics has only shipped 1500 robots in its entire lifetime and so when and and but at the same time we just had uh sonia on from sequoia capital a little bit ago and she's like very humanoid pill she was like i think they're coming like you know quickly years and so i'm curious as somebody who's actually building and shipping robots now kind of how you project out the next few years and i imagine uh you must be kind of entertained by it all it's it's gonna be you know there's a lot of capital on the line you know how hard it is but i'm curious how you think there there that's a great question there are two pieces of the puzzle there.

2:52:07One, I think you guys have touched in past interviews around accuracy and how accurate can robotics get. And the thing that we talk about internally is with AI today, we are collaborating. If it gets 90 % of right, we're pretty happy with it. With robotics, especially in a trivial task, we almost always want to delegate. We want to set it and forget. We don't want to do that last percent because I don't want to finish that last corner of the cleaning or that one last plate. and that actually puts bar much higher there is a corollary there as well which is we go to school maybe four years eight years learn how to do coding so if ai gets 90 percent of right we're mesmerized but we don't really go to school to learn how to navigate our home or how to pick up a glass or how to vacuum floors so the the trivial the task the more higher the accuracy expectations for customers because if you make mistakes in just picking up a glass they think of it as a dumb robot like come on yeah and then the analogy is imagine someone is helping you set a dining table um for your dinner party one out of hundred times and one out of hundred times they break one glass or one set of plate yeah there's a good chance you're gonna fire them so so the bar of accuracy is much higher there so that's one piece and then second thing is the adoption so this is where general magic is a good example in our mind general magic i tried to build iphone in 1995 5 didn't really work amazing team all x people 20 fidel was there and instead what we got was purpose-built device from cell phones to pdas to ipods to blackberries and then we combine everything into an iphone so in a similar way we tend to believe that purpose-built robots will see light of the day first and then they will get combined into multi-purpose devices and the more you have a human form the more expectations that customers would have so there's a home side of it and there is an enterprise out of it.

2:54:00So we tend to think that in enterprise or factories, there is a good chance robots would be used in two years. With homes, we think it's a little bit part of it. Yeah, that makes sense. Yeah, and if you're buying, let's say a humanoid comes in at even comparable to something like what Unitree is selling right now, and it's$40 ,000, your expectations on that are going to be gigantic. So it's going to be an uphill battle. But we'll have to have you back on as there's... Yeah, one last thing. What were we about to say? I was going to say you were touching on a really great point. We actually talk about it internally, that there is no ubiquitous consumer electronics device higher than$2 ,000.

2:54:39Yeah, TVs. Cars have been around for 100 years. Utility is clear. And that's usually$10 ,000,$20 ,000. And even then, it's a considered purchase. We just don't wake up and do it. So the utility and the value has to be proven. And then you have to convince customers to say, OKD is worth spending$10 ,000,$20 ,000, and it will survive five years, 10 years. So there is a productization element to the robotics that needs to be paid a little bit more attention to. Well, thanks so much for stopping by. This is a great chat. We will talk to you soon. We are going to use our Maddox. Thanks for having us.

2:55:11We're going to use our Maddox at the new studio. Yeah, we're excited. Absolutely. Tell us how it goes. We will. We'll let you know. We'll talk to you soon. Bye. Next up, we have Fastino coming in the studio. Sounds like an Italian name. Fastino. Fastino. ask him about it john you you love to you love to your italian accent is it's fantastic yeah it's my favorite uh big news festino trains ai models on cheap gaming gpus and just raised 17 and a half million dollars from kosla let's bring him into the studio we got uh greg here from festino or george george welcome how you doing guys how you doing you guys hear me all right yeah yeah uh first off uh how do you pronounce the company's name is it festino or fast so funny enough fastini in italian means a feast oh okay but fastino it's kind of a play on fast and tiny like very quick and small model it's our lame attempt at naming you've been feasting on uh nvidia graphics cards you've been feasting on cosla venture dollars uh putting them together can you break down uh what's the news this week all the above yeah yeah well great great being on guys that my last one of the week uh last one of the week tapping it off all right hoping to save the best for last you guys absolutely pretty fresh by the way i feel pretty underdressed yeah hit the soundboard let them know that we still got energy i want to hear the ashton hall effect not that one the other one let's go it's it's friday but we still got energy we're not we're not slacking off here next time i'm gonna have a suit please buy one please bring it on bring it on suits are underrated you should have one for every day of the week yes yes yes yeah i gotta go to new york in a couple weeks i think uh i gotta go first by the tailor we got a suit guy for you yeah yeah we'll introduce you anyway uh break down the news and then we'll we'll start uh talking about uh the business yeah so great being on uh this week we launched tlms so uh it's a family of language models called task specific language models.

2:57:15Yep. They're small, lightweight models that are really fast and they're built for AI developers. So being task specific really means that we're more accurate on enterprise tasks than large models like open AI or Gemini. And they cost a fraction to train. So we spent less than a hundred K on GPUs to train our models, but we're beating industry benchmarks for enterprise tasks. Okay. So are you fine tuning open source models? Are you ripping apart a mixture of experts model to just have a smaller set of weights? Give me the scope. I've seen the GPT 3.5 circles like this and the GPT four circles huge.

2:57:55How big is your circle? I guess. So we're not fine tuning or distilling any open source transformer based models. What we're doing is very different. We took a different approach from the large labs, we built a new architecture that maintains high accuracy, even with very low parameter counts. So we're not fully revealing exactly how many parameters, but all of our models are far below a billion parameters. Our architecture actually gets more accurate as the task becomes well-defined. So they're not generalist models. You cannot ask them to do anything, but they're extremely performant for the tasks that we built them for.

2:58:33Okay, talk about the data sources. I imagine that if you're doing summarization, you need a whole bunch of examples of that to train on. Text to JSON, probably need some text and some JSON. A lot of stuff's out there on the web. Are you scraping? Are you crawling? Are you buying data? Are you using open source data sets? Where's the data coming from? Yeah, really all of the above. I think there's a big debate out there in academia as to whether synthetic data or real world data leads to a more accurate model. We've definitely been using a blend of all the above, but the first models that we're rolling out and you nailed them perfectly.

2:59:09we have models for developers doing text to json text to sql we have an agentic function calling model that we're putting out i know you guys talk quite a lot about agents and yeah an agent that can book a flight for me or can book a hotel for me so we have a very lightweight model that inferences in milliseconds that can basically take what the user wants and call an api so very much developer focus we've got models that can parse documents redact private uh identifiable information from documents which is huge for banks insurance companies we have a really fun model it's our favorite model in-house it's a profanity censoring model i love that amazing we don't swear on the show so yeah yeah we try to keep it in the office but you can imagine we've had some fun late nights building synthetic data for profanity model which can be gaming can be used anywhere it's pretty much the funnest red teaming we can do i try to plug my ears or turn the air that's great uh i mean you're if you're inferencing this in milliseconds is there then a desire to uh deploy this at the edge you know run this in you know the cloud of the business that's actually deploying this so i just have to ask for the profanity model could we run it in real time while we're doing the show so if a guest you know drops a an f word f word Bleep.

3:00:32That's what I want. Yeah, it'll work in real time. It'll be much faster than an existing LLM. Yeah, I think the real TV shows, they have a system that puts it on delay and does something like this, but the delay has got to be so much faster if you're imagining like using Whisper and then this, there's a lot of things. But in terms of that latency, latency is really important to us. I imagine it's important to your clients and customers. Are you seeing demand for let us run your model and we'll still pay you, but we just don't want to go back and forth with your API. Yeah, so there are a lot of ways you can deploy smaller lightweight models.

3:01:06Obviously, we're going to be heavily reliant on our API. So when models have a small enough footprint, which actually comes from a lower parameter count, you can run them on-prem, you can run them on CPUs, low-end GPUs. I think Ash and I, my co-founder Ash, he had a dev agent, somewhat a cursor in 2023. After I sold my last company, I was actually an investor in his startup. He had a problem where his LLM costs ended up being higher than his headcount costs. And it's a problem that a lot of agent companies face. So I was actually, I spent a little bit of time as a GP after I sold my last company and all of our portfolio companies were facing the same thing.

3:01:47Rising costs of LLM. So it's not only latency, as you mentioned, accuracy is a big problem with large LLMs. But frankly, the key issue that we've seen is that LLMs just aren't built for the enterprise. They're built for consumers, right? So GPT, Gemini, they're trained on trillions of data points. They're used by our friends, our family every day. They help you code. They help you get food recipes. They help you prep for podcast interviews. They're not built for high-scale enterprise tasks, right but enterprises are spending millions of dollars a month on these large monolithic apis i want to tell you guys a story sure how gbt is being used so my my wife's dog got really sick a few weeks ago um cute little guy 13 years old he has cancer we went to the pet hospital and the doctor basically recommended that we put him down we had her best friend on the phone and we're trying to make a decision based on the doctor's recommendation, symptoms, dog's age, what to do.

3:02:52And who is the tiebreaker? We asked GBT. Hey, GBT, here's what's going on with my dog. Here's the situation. Should we put him down, right? So my wife was using GBT to play maker. And for the record, GBT told us to put him down and we didn't listen. He's doing well today. But wow, that's a crazy story. Why would AI is the antichrist? Are trying to take out the dot, the poor pup. Yeah. Why would a large bank, Bank of America, city, JPMC, if all they're looking to do is analyze your bank statement or look at some log for fraud, why are they using the same model model that my wife used to consult on her dog's mortality?

3:03:40Right. So using these massive models, it's like trying to come up with a cool metaphor for the show. It's like the DoorDash guy riding Saturn V to come in for a pizza, right? It's so unnecessary. And frankly, that's why large enterprises, large banks haven't put ChatGPT or an LLM into a chatbot. Yeah. Super fascinating. Analyzing your bank statement and it's just like$500 on dinner? You cannot afford to have a dog. You got to put that dog down. It's like, Chase, what are you doing? What are you doing? JP Morgan, cool it with the recommendations. Reasoning is a very sexy word in this space right now.

3:04:20But from speaking with almost 100 Fortune 500 enterprises since we announced our pre-seed round in Q4, they don't want models that can reason. A bank does not want a model that can reason its way through your last 100 chatbot users logs and figure out their personal information. So when you're building lightweight task-specific models like this, our models are in domain. They're only trained to do the task that the enterprise is using them for. So I think we have that edge very much just in how the models are built. Okay. Talk about the gaming GPUs worth less than$100 ,000 in total is what TechCrunch is reporting.

3:04:59Did you build that yourself? Do you have your own data center? or are there clusters out there of low end GPUs that are all rigged together from like legacy Bitcoin mining applications or something like that? Where are you getting these? Yeah. So we have GPUs in house. We have, we use GPUs in the cloud, but we all told our models take about a couple hours to train. The training costs for one model is less than the cost of a Chipotle burrito i think so much said about how but a 2012 chipotle burrito or 2023 one that didn't give me food poisoning hopefully but there there's been so much said about when deep seek came out and they only spent 10 million dollars on h100s however questionable that number was i think what we're trying to do as a super small team is show that you don't even need h100s to build generative models for enterprises right so we didn't use a single h100 we used t4s gaming gpus low-end v100s and and you can do that uh i think we've proven that you know banks don't need a model that takes six months to run or it's going to drain you know lake tahoe to for a training run right so what's more important with that training run uh flops or memory because i imagine lower token counts you haven't released it but i imagine you can fit it in memory and so that unlocks it but um talk to me about the dynamics of like building a cluster and thinking about the the different parameters that go into the cards that you select yeah so we have a family of of it's going to be less than 10 models most likely for the next six months i'll say that they're far below a billion parameters each.

3:06:51So we don't need a cluster even. We can just run these on one or two GPUs each. For inference? For inference. Yeah. And for training, it's the same thing. It's a very low end GPU for an hour or two. So we definitely believe that there's going to be a giant shift in how language models are used. Right. So you guys have seen waves of the last couple of decades. So these massive IBM mainframes shifted into client server architectures and open source software. You used to have these massive monolithic builds that would take, you know, that would ship a month at a time. Software applications were shipped so much slower.

3:07:32And then out came microservices and you have a different release for your payment gateway and your APIs. This kind of workload partitioning, as we call it, it's completely going to change the landscape in language modeling. So there was a report from Gartner that came out about a month ago saying that small task specific models are going to outpace LLMs and enterprise usage by three to one in three years. And we want to lead that. We definitely think that every developer is going to become an AI developer. So every dev today will need to be able to integrate a language model into their code, just like they're integrating an open source NPM package or Python library.

3:08:15it needs to be much simpler and that's very much how we're looking to change the game it's going to be really hard to compete with the big labs yeah if we're just focused on the models which obviously we're a foundational model company yeah but we need to make much smoother developer workflow integrations we need to make life easier for devs and right now it's yeah where does this yeah where does this go? I mean, I know some of the stories about companies like Ramp, for example, our sponsor, they need to digitize receipts. So they get a lot of images. They do OCR. And I think Google provides an API for that.

3:08:55There's a bunch of companies that do OCR. It comes through as kind of messy cluster of text. Then pipe that through GPT-4 and boom, you have structured data. Lama comes out. okay, maybe it's getting cheaper, but this seems like something where you'd want to go to you guys and get an even cheaper model that's distilled just for that one task. But that feels like almost like you're a consulting shop or is there a place where a company says, hey, we've been using GPT-4 or LLAMA and we've done 10 million inferences. And so we have a lot of data about what works, what doesn't. Can you train a custom model for us to drop our inference cost by a couple orders of magnitude?

3:09:37Or are you trying to focus more on more versatile foundation models that can be just tools in the tool chest and aren't kind of one-off specific systems for a specific task within a specific company? Yeah, I think there are a few ways to answer that. The first one is probably talking about agents and how agentic systems are evolving. I'm definitely in the camp of thought that says that agentic systems will take over legacy SaaS systems within four to five years right so when you see how these agents are being put together it's typically daisy chaining eight or ten llm calls so in a chat bot you want to parse a query you want to then figure out the right document to give back to the user summarize the right chunk of that document give it back all in real time uh with a very smooth chat interface so we definitely see a world where you have different models for different tasks.

3:10:33And we're not saying that we're going to replace large generalist LLMs. The models that are larger, they're good at reasoning, they're good at research, they do things like orchestration. So they'll help orchestrate this entire pipeline. That's still going to be the large model. That's still going to be your GPTs. You're going to be your Gemini's. But the actual agents, the workers that are calling APIs, that are doing these sort of deterministic, high-scale, high-throughput tasks, those are all going to be very small, intelligent, task-specific models. That's how we see it play out. NVIDIA is down 0.6 % today.

3:11:15Is that because of you? Is the market pricing? I mean, billions of dollars. Billions of dollars have been evaporated from the markets on the news. How do you think, assuming Festino just gains massive adoption over the coming years, how do you think that impacts the GPU market, GPU demand broadly? Yeah. For the record, the low-end GPUs that we use are still NVIDIA GPUs. We're still a big NVIDIA. Another$10 trillion to Jensen. Jensen might appreciate if you weren't so efficient and you raised$500 million and gave it to him. Have you thought about doing that? If he can help us scale and go more viral, we'll need more of them for inference for the record.

3:12:02But I think there's going to be a huge demand for consumers, for LLMs as they are. I think we were lucky enough to have two of the first OpenAI investors on our cap table. That's right. In the very beginning, when there was a pitch deck, there wasn't a business model. They didn't think that they were going to be a consumer company. But the consumer appetite for LLMs has gone crazy. Even during the DeepSeek moment, didn't DeepSeek get to number one on the App Store? My wife, my friends are all downloading DeepSeek. So the need and the hunger for consumers to automate their lives will constantly be driving the need for these larger generalist LLMs.

3:12:44We just don't believe they're needed for the enterprise. So we're taking a very different approach. Very cool. Makes total sense. Anything else? I love this conversation. Me too. This is super fun. He's an absolute dog. You're an absolute dog. You're an absolute dog. Call me anytime you guys want to talk about LLMs. And I might need to get some help on buying a suit for New York. Oh, fantastic. We've got you. And seriously, we... We got you. You got us. It's one hand. I want to figure out this real-time censoring thing. I think it'd be hilarious. We have some guests that come on and try to drop F-bombs.

3:13:14It's unacceptable. Our children listen to this. Yeah, it's unacceptable. And I think it'd be hilarious if it was like, you know, made like a duck sound. Yeah, yeah, quack. That'd be great. So our people will talk to your people. Yeah, appreciate it, guys. Thanks so much for having us on. Thank you so much. Hey, big news. Oh, yeah, yeah. Rippling has raised$450 million at a$16.8 billion valuation. Huge news. And bigger news. Oh, I was actually supposed to be texting with them. Let me see. YC is a customer. Really? And that is big considering YC has created all the big payroll companies. Gusto, Deal, Rippling.

3:13:56Gary must have been finding it hard to pick favorites. Well, I'm texting with the team. Hopefully get Parker on soon to talk about the business. One last thing. We have a couple of posts we want to go through. But Michael, Can you check the printer? Because I think we got a special printout for today. We haven't been printing posts very frequently, but we got one post that I wanted to print. Hopefully it printed. Let me see. I don't even know if it printed. It did not print. It did not print. All right, what's the post, John? Is it working? I'm trying to print it again. and we might be out of paper, but we can pull up the digital version.

3:14:46We're going to run it back. Anyway, Luke Metro said, this show used to print out my tweets and read them. Now they have the head of the army. So thank you to Luke Metro. I tried to print it, but we haven't printed in so long that the printer's not working. Anyway, there's some other posts. Did you see the drama in Anderil World about Matt Grimm taking notice.co to task for selling some fake equity in the company. Yeah. So it's hard to know. I mean, the way that notice was displaying this information was not consistently candid is how I would describe it. And then the funniest thing, you called this out to Matt, apparently the CEO of notice, messaged Matt and said, Hey, I'm FINRA registered, so not supposed to post publicly on social media.

3:15:41Happy to continue the convo privately or do a call if you want. Let me know, buddy. Buddy. It's a war crime. Hitting Matt Graham with buddy is a war crime. Straight to jail. It's so bad. It's so brutal. It's so bad. I actually think the Notice platform is pretty cool. Oh, yeah. They have a bunch of, you know. I mean, I imagine that there's probably some steel man here. There's a million ways that, you know, it's a big company. There's a lot of investors. Someone could have come and figured out a way to put some shares on there. No, I think what was happening is an employee brought shares. Oh, really?

3:16:18Okay. Like the reason that the way that you would have zero fees is that an employee. Sure, is if you own it. It's common. But it was, yeah, it's common. And that wasn't being necessarily disclosed. And there should be transfer restrictions. I mean, these are very standard. So there's something odd. I mean, there's been a big, big history of these odd, like, secondary sales. like uh for a while people were doing forward contracts did you ever follow that story oh yeah so basically i mean i think they're still work i i think they're definitely banned in most companies you're not supposed to do a contract and it's written in your in your but the whole nature of a forward contract is that the company doesn't really there's very good chance the company would never find out true but they can still ban it in your employment agreement like it can still be something that you agree to when you join the company.

3:17:03So basically a forward contract, if you're not familiar, it's the right to purchase the shares at a future date, much like a stock option. So you're basically writing an options contract against your shares in the third party. And the shares themselves don't actually transfer. And so the company, in theory, doesn't need to approve it, but it's very standard. And they wouldn't necessarily know. And they wouldn't necessarily know. Yeah. And so this was a big thing, very, very controversial. a lot of people don't like this. And so even though it doesn't result in a whole bunch of like legal complications, obviously you're not transferring the information rights because you're not actually transferring the shares.

3:17:38They couldn't sue the company. That's the big reason why you want a clean cap table is because you have to deal with every investor. You have to give them information. They can sue you. If they just own a forward contract, they probably can't do that. But it's still a problem and creates all these distortions in the market. And so Matt Grimm taking them to task. He says, we're one of the good guys. as CEO of a company who solicits retail investors to buy shares at a significantly inflated price for what it's worth in a privately held company they do not own shares of, have direct access to, or have any information or information rights from which to discern financial performance or market positioning or anything whatsoever to inform the proposed investment, all while not clarifying publicly, or for that matter, to the clients they are soliciting, how exactly this exposure is structured or what precisely these clients are buying.

3:18:23And while privately hiding behind a claimed veil fake by the way their finra registration means they can't comment when keeping it real goes too far uh rough so clean it up you don't want to have that grim on your bad side he could basically say this about most secondary brokers and platforms anyway we got some massive breaking news we got some personnel jacob afron's been promoted to managing director at Redpoint. We love to see it. Very interesting. The rumors, this is a maxed out contract. Rumors have been swirling for a long time. He was in free agency, considered free agency, stuck with Redpoint, got promoted.

3:19:04Congratulations to him. Couldn't be more excited for everything I had. And weird terminology over at Redpoint. Not GP, not partner, managing director. Very investment bank-like, you know? Official business. Andrew Reid had a funny post he's a uh coming from banking no matter how senior i get in venture i'll always think that a managing director is uh more senior to me mogged feels feels very senior it does it does um anyway uh arvin srinivas over at perplexity is taking more shots at bloomberg bloomberg is such a joke while perplexity does real-time call transcriptions for free bloomberg has a 15 minute lag and it costs$30 ,000 a year.

3:19:45To be clear, that was Marcello who said that. Somebody at TechCrunch will listen to this and say, Arvind said Bloomberg is a joke. Anyways, what perplexity is doing around real-time call transcription. Imagine if with TechCrunch reports is my quote. Me quoting someone else as Arvind. That's what I was saying. That's so bad. That's a risk. The bar's low these days, folks. The bar's low. So anyway, any other posts we want to go through? The Pope is American. Vatican City has a Buc-ee's now and a Waffle House and a Costco. It's great. They've taken over. America is in control. It's fantastic. America is in control.

3:20:26We hope you have a great Mother's Day. Pick up something simple. Thank you for joining us this week, folks. We had a great time doing this show. Yeah, it was fantastic. It was a good job. Thank you for tuning in. And we'll see you next week. And remember, it's Mother's Day. We talked about this three hours ago. The Super Bowl of pronatalism, folks. Take care, everybody. Have a great Mother's Day. Cheers.